Consciousness and explanation - mkdavies.net · Consciousness and explanation ... tions and...

Chapter 1

Consciousness and explanation

Martin Davies

1.1 Two questions about consciousness: ‘what?’ and‘why?’

Many aspects of our mental lives are conscious—an ache in tired muscles; thesight, smell, and taste of a glass of wine; feelings of happiness, love, anxiety orfear; trying to work out how best to test a hypothesis or structure an argument.It seems beyond dispute that at least some sensations, perceptions, emotionalepisodes, and bouts of thinking are conscious. But equally, there is much inour mental lives that is not conscious. It is a central idea in cognitive sciencethat there can be unconscious information processing. It is also plausible thatthere can be unconscious thought and unconscious emotions; there are casesof ‘perception without awareness’; and perhaps even bodily sensations cansometimes be unconscious.1 What, then, is the difference between consciousand unconscious mental states? Is there, for example, something distinctiveabout the neural underpinnings of conscious mental states? An answer to this‘what?’ question could be called (in some sense) an explanation of consciousness.

We might, however, expect rather more from an explanation of consciousnessthan just a principle or criterion that sorts conscious mental states fromunconscious ones. Suppose that we were told about a neural condition, NC, thatwas met by conscious mental states but not by unconscious ones. Suppose thatthis was not just an accidental correlation. Suppose that the difference betweenmeeting this neural condition and not meeting it really was the difference that

1 Claims about unconscious thoughts and emotions are common in, but not restricted to,the psychoanalytic tradition. In ordinary life, it sometimes seems that we arrive at a solu-tion to a problem by processes of thinking that do not themselves surface inconsciousness, although their product does. For the conception of emotion systems asunconscious processing systems whose products are sometimes, but not always, availableto consciousness, see LeDoux (1996, this volume). The term ‘perception without aware-ness’ is applied to a wide range of phenomena (Merikle et al. 2001) including blindsight(Weiskrantz 1986, 1997). For the proposal that unconscious mental states may includeeven sensations, such as pains, see Rosenthal (1991, 2005); for a recent discussion, seeBurge (2007, pp. 414–419; see also 1997, p. 432).

01-Weiskrantz-Chap01 7/8/08 11:17 AM

makes the difference. There would remain the question why mental states thatmeet condition NC are conscious. Even if condition NC were to make the difference, it would not be a priori that it makes the difference. It would seemperfectly conceivable that condition NC might have been met in the absence ofconsciousness. So—we would ask—why, in reality, in the world as it actually is,is this condition sufficient for consciousness? The problem with this ‘why?’question is that, once it is allowed as legitimate, it is apt to seem unanswerable.

The intractability of the ‘why?’ question is related to our conception of con-sciousness, a conception that is grounded in the fact that we ourselves aresubjects of conscious mental states. The situation would be quite different ifour conception of consciousness were a third-person conception, exhausted bystructure and function—if it were a physical-functional conception. Our con-ception of a neurotransmitter, for example, is a physical-functional conception.There is a ‘what?’ question about neurotransmission: What chemicals makethe difference? But, once we know the structure and function of GABA ordopamine, its role in relaying, amplifying, and modulating electrical signalsbetween neurons, there is no further question why it is a neurotransmitter.That is just what being a neurotransmitter means. Similarly, if our conceptionof consciousness were a physical-functional conception then lessons about thenature of condition NC and about its role in the overall neural economy,about its constitution and connectivity, could persuade us that neural condi-tion NC was consciousness—or, at least, that NC played the consciousnessrole in humans—because it had the right structure and function.

As things are, however, our conception of consciousness does not seem to beexhausted by structure and function and the ‘why?’ question remains. A neu-roscientific answer to the ‘what?’ question would be of great interest but it wouldnot render consciousness intelligible in neuroscientific terms. Consciousnesswould remain a brute fact. Between neural (or, more generally, physical) condi-tions and consciousness there is an explanatory gap (Levine 1983).

1.1.1 Positions in the philosophy of consciousnessSometimes, the explanatory gap is presented as licensing a conclusion aboutthe nature of reality itself, and not just about our conceptions of reality. It isargued that the existence of an explanatory gap between the physical sciencesand consciousness supports the conclusion that consciousness is metaphysi-cally or ontologically distinct from the world that the physical sciences describe.It would be no wonder that consciousness could not be explained in terms ofthe physical sciences if consciousness were something quite different from thephysical world. The conclusion that consciousness falls outside the physicalorder is sometimes dramatized as the claim that there could, in principle, be a

CONSCIOUSNESS AND EXPLANATION2


creature physically just like one of us yet lacking consciousness—a zombie(Chalmers 1996; Kirk 2006), or even a complete physical duplicate of our worldfrom which consciousness was totally absent—a zombie world. In line with thisclaim, David Chalmers proposes that ‘a theory of consciousness requires theaddition of something fundamental to our ontology, as everything in physicaltheory is compatible with the absence of consciousness’ (1995, p. 210).

While Chalmers argues that consciousness is not wholly physical, it is morecommon (at least within academic philosophy) to assume or argue that someversion of physicalism is true, so that consciousness must be part of the physi-cal world (Papineau 2002).2 Contemporary physicalists reject the duality ofmaterial and mental substances that Descartes proposed and also reject theduality of material and mental properties or attributes. According to physical-ism, conscious mental states, processes, and events are identical to physical(specifically, neural) states, processes, and events. Furthermore, the phenome-nal properties of conscious mental states (what being in those states is like forthe subject) are the very same properties as physical properties of neural statesor—if the claim of identity between phenomenal and physical propertiesseems too bold—the phenomenal properties are strongly determined by phys-ical properties. The idea of strong determination in play here is that thephenomenal properties are necessitated by the physical properties. The phe-nomenal properties do not and could not vary independently of the physicalproperties; they supervene on the physical properties.3

Physicalist approaches to the philosophy of consciousness come in two varieties. Chalmers (1996) calls the two kinds of approach type-A materialismand type-B materialism. Some physicalists (type-A materialists) deny thatthere is an explanatory gap and maintain, instead, that consciousness can befully and satisfyingly explained in physical terms. This option is, of course,mandatory for physicalists who agree with anti-physicalists like Chalmers thatthere is a good argument from the existence of an explanatory gap to the con-clusion that consciousness falls outside the physical order.

Other physicalists (type-B materialists) allow that there is an explanatorygap but deny that there is a good argument from the gap to the anti-physicalistconclusion. In his development of the notion of an explanatory gap, JosephLevine (1993) distinguishes two senses in which it might be said that the physical

TWO QUESTIONS ABOUT CONSCIOUSNESS: ‘WHAT?’ AND ‘WHY?’ 3

2 See Zeman (this volume, section 11.2.3) for some data about public understanding of themind. In a survey of undergraduate students, ‘64% disputed the statement that “the mindis fundamentally physical”’ (p. 294).

3 We shall return (section 1.7.2) to the distinction between the strict version of physicalism(phenomenal properties are identical to physical properties) and the relaxed version(phenomenal properties are determined by, or supervene on, physical properties).


sciences leave out consciousness, the epistemological sense and the metaphysi-cal sense. The claim that the physical sciences leave out consciousness in theepistemological sense is the claim that there is an explanatory gap. The claimthat the physical sciences leave out consciousness in the metaphysical sense isthe claim that consciousness falls outside the physical order. Levine says thatthe distinction between epistemological leaving out and metaphysical leavingout ‘opens a space for the physicalist hypothesis’ (1993, p. 126). Type-B mate-rialist approaches typically involve two claims. First, the explanatory gapresults from our distinctively subjective conception of consciousness. Second,there can be both scientific conceptions and subjective conceptions of thesame physical reality (just as, in the familiar case of Hesperus and Phosphorus,there can be two concepts of a single object, the planet Venus). Type-B materi-alists maintain that there can be a duality of conceptions without a duality ofproperties.

1.1.2 OutlineThis chapter begins with the subjective conception of consciousness that givesrise to the explanatory gap and the intractability of the ‘why?’ question. Next,there is a discussion of the approach to the study of consciousness that wasadopted by Brian Farrell (1950), an approach that frankly rejects the subjectiveconception in favour of a broadly behaviourist one.4 Farrell’s approach servesas a model for subsequent type-A materialists.

The second half of the chapter is organized around Frank Jackson’s knowledgeargument—an argument for the anti-physicalist claim that phenomenal proper-ties of conscious mental states are not physical properties. The knowledgeargument ‘is one of the most discussed arguments against physicalism’ (Nida-Rümelin 2002) and the philosophical literature of the last 25 years contains manyphysicalist responses to the argument. Perhaps the most striking response isJackson’s own, for he now rejects the knowledge argument, adopting a type-Amaterialist approach and denying that there is an explanatory gap.

The type-A materialism that Jackson shares with Farrell denies that there isanything answering to our conception of consciousness to the extent that the conception goes beyond structure and function. In that respect, type-Amaterialism ‘appears to deny the manifest’ (Chalmers 2002, p. 251), and isprobably the minority approach amongst philosophers who defend physicalism


4 Brian Farrell was Wilde Reader in Mental Philosophy in the University of Oxford from1947 to 1979. He died in August 2005, at the age of 93. The institutional and historical set-ting of the lecture on which this chapter is based (Oxford in the spring of 2006) invitedextended reflection on Farrell’s paper.


(although it is the approach adopted by such influential figures as DanielDennett and David Lewis). The more popular approach is type-B materialism,accepting that there is an explanatory gap but denying that this leads to theanti-physicalist conclusion (Chalmers 1999, p. 476): ‘It simultaneously prom-ises to take consciousness seriously (avoiding the deflationary excesses oftype-A materialists) and to save materialism (avoiding the ontological excessesof the property dualist).’ By considering the knowledge argument and responsesto it, we shall be in a position to assess the costs and benefits of some of themost important positions in contemporary philosophy of consciousness.

1.2 The subjective conception of consciousnessWe have distinguished two questions about the explanation of consciousness,the ‘what?’ question and the ‘why?’ question. The question what makes the dif-ference between conscious and unconscious mental states seems to be atractable question and many scientists and philosophers expect an answer inbroadly neuroscientific terms—an answer that specifies the neural correlates ofconsciousness (Chalmers 2000; Block 2005; Lau, this volume). The questionwhy this neuroscientific difference makes the difference between conscious andunconscious mental states is more problematic. As Thomas Nagel put the pointover 30 years ago, in his paper, ‘What is it like to be a bat?’ (1974/1997, p. 524):

If mental processes are indeed physical processes, then there is something that it islike, intrinsically, to undergo certain physical processes. What it is for such a thing tobe the case remains a mystery.

Ned Block expressed a similar view in terms of qualia, the subjective,phenomenal, or ‘what it is like’ properties of conscious mental states (1978,p. 293):

No physical mechanism seems very intuitively plausible as a seat of qualia, least of all abrain. ... Since we know that we are brain-headed systems, and that we have qualia, weknow that brain-headed systems can have qualia. [But] we have no theory of qualiawhich explains how this is possible.

1.2.1 Nagel’s distinction: subjective and objectiveconceptions

Nagel’s announcement of mystery was not based on gratuitous pessimismabout the progress of science but on an argument. The starting point was thethought that we cannot conceive what it is like to be a bat. The conclusion wasthat, although we can (of course) conceive what it is like to be a human, wecannot explain, understand, or account for (our) conscious mental states interms of the physical operation of (our) brains. We should take a moment to

THE SUBJECTIVE CONCEPTION OF CONSCIOUSNESS 5


review the steps that led Nagel from the alien character of bat consciousness tothe mystery of human consciousness.

The initial thought about bats can be extended to a distinction between twotypes of conception. We cannot conceive what it is like to be a bat and likewisea bat or a Martian, however intelligent, could not conceive what it is like to bea human—for example, what it is like to undergo the conscious mental statesthat you are undergoing now. These limitations reflect the fact that concep-tions of conscious mental states as such are subjective; they are available fromsome, but not all, points of view. Roughly, the conscious mental states that wecan conceive are limited to relatively modest imaginative extensions from theconscious mental states that we ourselves undergo. We cannot conceive what itis like to be a bat although we can conceive what it is like to be human. Wecannot conceive what it is like for a bat to experience the world throughecholocation although we can conceive what it is like for a human being toexperience the red of a rose or a ripe tomato.

While grasping what a conscious mental state is like involves deployment ofsubjective conceptions, the physical sciences aim at objectivity in the sensethat the conceptions deployed in grasping theories in physics, chemistry, biology,or neuroscience are accessible from many different points of view. The physicaltheories that we can grasp are limited, not by our sensory experience, but byour intellectual powers; and the conceptions that are required are, in principle,no less available to sufficiently intelligent bats and Martians than to humans.

1.2.2 Knowing what it is likeNagel said (1974/1997, p. 521; emphasis added), ‘I want to know what it is likefor a bat to be a bat’, and he went on to point out that the expression ‘knowingwhat it is like’ has two different, though related, uses (ibid. p. 526, n. 8; see alsoNida-Rümelin 2002, section 3.3). In one use, knowing what a particular typeof experience is like is having a subjective conception of that type of experience.There is a partial analogy between knowing what a type of experience is likeand the ‘knowing which’ that is required for thought about particular objects(Evans 1982).5 Knowing what a type of experience is like is similar to knowing


5 Gareth Evans’s (1982) theorizing about object-directed thoughts was guided by Russell’sPrinciple, which says that in order to think about a particular object a thinker must knowwhich object it is that is in question. Evans interpreted the principle as requiring discrim-inating knowledge, that is, the capacity to discriminate the object of thought from allother things. Initially, this may sound so demanding as to make object-directed thoughtan extraordinary achievement. But Evans’s examples of ways of meeting the ‘knowingwhich’ requirement make it seem more tractable.


which object is in question in being a kind of discriminatory knowledge. Inthe case of thought about particular objects, there are many ways of meetingthe ‘knowing which’ requirement: for example, presently perceiving the object,being able to recognize it, or knowing discriminating facts about it. In the caseof thought about types of experience, there may also be many ways of meetingthe ‘knowing which’ requirement. Having a subjective conception of a type ofexperience is meeting the ‘knowing which’ requirement in virtue (roughly) ofbeing the subject of an experience of the type in question. (We shall refine thisshortly.)

In a second use, knowing what it is like is having propositional knowledgeabout a type of experience, conceived subjectively. It is not easy to provide aphilosophical account of having a conception or concept, but a subject whohas a conception of something has a cognitive capacity to think about thatthing. A subject who has a conception of a type of experience can deploy thatconception in propositional thinking and may achieve propositional knowl-edge about that type of experience. He might know that he himself is havingan experience of that type, or that he has previously had such an experience;and he may know something of the circumstances in which other people haveexperiences of that type. In the latter case, the subject knows what it is like forpeople to be in those circumstances.

It is plausible that a subjective conception of a type of experience can bedeployed in thought even when the subject is not having an experience of thetype in question. If that is right, then it must be possible for a subject to meetthe ‘knowing which’ requirement in respect of a type of experience withoutconcurrently being the subject of an experience of that type. On some accountsof having a subjective conception, it might be that remembering being thesubject of an experience of the type in question would be sufficient to meetthe ‘knowing which’ requirement. (Perhaps having a veridical apparentmemory would suffice.) Alternatively, it might be proposed that meeting the‘knowing which’ requirement involves being able to imagine being the subjectof an experience of the type in question or being able to recognize other tokenexperiences of which one is the subject as being of the same type again. (We shallreturn to these abilities to remember, imagine, and recognize in section 1.10.2.)

Michael Tye (2000) suggests that there are two different ways in which asubject can meet the requirements for having a subjective conception of a typeof experience. In the case of a relatively coarse-grained experience type, suchas the experience of red, a subject might meet the ‘knowing which’ require-ment on the basis of long-standing abilities to remember, imagine, andrecognize experiences of that type. In the case of a very fine-grained experi-ence type, such as the experience of a specific shade of red, the limitations of



human memory may prevent a subject from reliably discriminating laterexperiences of that precise type from others. Nevertheless, it seems that a subjectwho is actually having an experience of that shade of red (and whose attentionis not occupied elsewhere) has a subjective conception of that fine-grainedexperience type and knows what it is like to experience that specific shade ofred. In such a case, the subject meets the ‘knowing which’ requirement invirtue of being the subject of an experience of the fine-grained type in ques-tion even if possession of the subjective conception lasts no longer than theexperience itself.

1.2.3 Nagel’s conclusion: physical theories and theexplanation of consciousness

With these two uses of ‘knowing what it is like’ in mind, we can distinguishtwo claims that are immensely plausible in the light of Nagel’s distinctionbetween subjective and objective conceptions. The first claim is that subjectiveconceptions cannot be constructed from (are not woven out of) the objectiveconceptions that are deployed in grasping theories in the physical sciences.A subject might be able to deploy all the objective conceptions needed to graspphysical theories about colour vision without having any subjective concep-tion of the experience of red. The second claim that is plausible in the light ofNagel’s distinction is that there is no a priori entailment from physical truthsto truths about conscious mental states conceived subjectively.

The second claim is not an immediate consequence of the first (Stoljar 2005;Byrne 2006) because a priori entailment of subjective truths by physical truthsdoes not require that subjective conceptions should be constructible fromphysical conceptions. The second claim says that a subject who was able todeploy objective conceptions of physical states and who also possessed the sub-jective conception of a particular type of experience would not, just in virtue ofhaving those conceptions, be in a position to know that a person in such-and-such a physical state in such-and-such a physical world would have anexperience of that particular type.

If these claims are correct then physical theories, to the extent that theyachieve the objectivity to which science aspires, will not say anything aboutconscious mental states conceived subjectively. We know what it is like toundergo various conscious mental states, but the conceptions that constituteor figure in that knowledge have no place in our grasp of objective physicaltheory. Nor will the content of our distinctively subjective propositionalknowledge about conscious experience be entailed a priori by physical theory.

Once we grant the contrast between subjective conceptions and the objec-tive conceptions that are deployed in grasping physical theories, the conclusion



of Nagel’s argument is compelling. We cannot explain conscious mental statesas such—that is, conceived subjectively—in terms of the physical operation ofbrains conceived objectively.

In a similar spirit to the Nagelian argument, Colin McGinn says (2004, p. 12):

any solution to the mind-body problem has to exhibit consciousness as conservativelyemergent on brain processes: that is, we must be able to explain how consciousnessemerges from the brain in such a way that the emergence is not radical or brute.

And (ibid., p. 15):

What the theory has to do is specify some property of the brain from which it followsa priori that there is an associated consciousness …. A priori entailments are whatwould do the trick.

But a priori or conceptual entailments will not be available precisely becauseof the ‘vastly different concepts’ (p. 19) that figure, on the one hand, in thephysical sciences of the brain and, on the other hand, in our knowledge ofwhat it is like to undergo conscious mental states.

1.2.4 Subjective conceptions and physicalismAccording to Nagel’s argument, the explanatory gap is a consequence of thedistinction between subjective conceptions and the objective conceptions thatare deployed in grasping physical theories. On the face of it, this duality ofconceptions is consistent with the truth of physicalism and, indeed, at the endof his paper, Nagel says (1974/1997, p. 524): ‘It would be a mistake to concludethat physicalism must be false.’

If physicalism is true and conscious mental states fall within the physicalorder then they are part of the subject matter of objective physical theory.Similarly, if thinking about things, or conceiving of things, falls within thephysical order then the activity of deploying conceptions—even deploying sub-jective conceptions—is part of the subject matter of objective physical theory.Thus, when we grasp physical theories by deploying objective conceptions,we may think about a physical event or process that is, in fact, the deploymentof a subjective conception. But this does not require us to be in a position,nor does it put us into a position, to deploy that subjective conception our-selves. Even on a physicalist view of what there is in the world, graspingphysical theories is one thing and deploying subjective conceptions is another.(In sections 1.11 and 1.12, we shall consider arguments that this duality ofconceptions is not, in fact, consistent with physicalism.)

Tye argues that the explanatory gap presents no threat to physicalismbecause, really, there is no gap (1999/2000, p. 23): ‘it is a cognitive illusion’.By claiming that there is no gap, Tye does not mean that there really are a priori



entailments from physical truths to truths about conscious mental states conceived subjectively. He agrees with Nagel that the distinction between sub-jective and objective conceptions guarantees that there are no suchentailments. But he argues that it is a mistake to describe the absence of suchentailments as a gap (ibid., p. 34):

[T]he character of phenomenal [subjective] concepts and the way they differ fromthird-person [objective] concepts conceptually guarantees that the question [why it isthat to be in physical state P is thereby to have a feeling with this phenomenal character]has no answer. But if it is a conceptual truth that the question can’t be answered, thenthere can’t be an explanation of the relevant sort, whatever the future brings. Since anexplanatory gap exists only if there is something unexplained that needs explaining,and something needs explaining only if it can be explained (whether or not it lieswithin the power of human beings to explain it), there is again no gap.

There are at least two important points to take from this bracing passage.First, if there are distinctively subjective conceptions of types of experiencethen there will be truths about conscious experience that are not entailed a prioriby physical truths. So, a philosopher who maintains that all truths about con-scious experience are entailed a priori by physical truths (a type-A materialist)must deny that there are distinctively subjective conceptions of the kind thatNagel envisages. Second, the absence of a priori entailment from physicaltruths to truths about conscious experience (subjectively conceived) is con-ceptually guaranteed (Sturgeon 1994). So, it is not an absence that will beovercome by progress in the physical sciences.

I shall not, myself, put these important points in Tye’s way. Instead of saying,with Tye, that there is no explanatory gap, I shall say that there is an explana-tory gap if there is no a priori entailment from physical truths to truths aboutconscious mental states conceived subjectively. The difference from Tye is ter-minological. I am prepared to allow that an explanatory gap exists eventhough what is unexplained is something which, as a matter of conceptualtruth, cannot be explained.

1.3 Farrell on behaviour and experience: Martians and robots

In discussions of Nagel’s (1974) paper, it is often noted that the ‘what it is like’terminology and, indeed, the example of the bat, occurred in a paper by BrianFarrell, ‘Experience’, published in the journal Mind in 1950. I shall come in amoment to the use that Farrell made of the bat example. Before that,I need to describe the problem that Farrell was addressing—a problem which,he said, troubled physiologists and psychologists, even if not ‘puzzle-wise pro-fessional philosophers’ (1950, p. 174).



The problem is that scientific accounts of ‘what happens when we think,recognize things, remember, and see things’ leave something out, namely,the experiences, sensations, and feelings that the subject is having (ibid.,p. 171).6 The experimental psychologist, for example, gathers data about asubject’s ‘responses and discriminations’, dealing with ‘behaviour’ but not with‘experience’. Thus (p. 173): ‘while psychology purports to be the scientificstudy of experience,… the science, in effect, does not include experiencewithin its purview’. The problem that troubled the physiologists and psycholo-gists was, in short, that the sciences of the mind leave out consciousness.

Farrell argues that there is really no such problem as the physiologists andpsychologists take themselves to face. He asks us to consider the sentence(1950, p. 175):

If we merely consider all the differential responses and readinesses, and such like, thatX exhibits towards the stimulus of a red shape, we are leaving out the experience hehas when he looks at it.

He argues that this is quite unlike ordinary remarks, such as:

If you merely consider what Y says and does, you leave out what he really feels behindthat inscrutable face of his.

The difference between the two cases is said to be this (p. 176): ‘What we leaveout [in the second sentence] is something that Y can tell us about [whereas]what is left out [in the first sentence] is something that X cannot in principletell us about’. But why is it that X cannot tell us about what seems to be left outby a description of responses and readinesses, namely, his experience? Farrellanswers (ibid.):

He has already given us a lengthy verbal report, but we say that this is not enough. Wewant to include something over and above this, viz., X’s experience. It is useless to askX to give us further reports and to make further discriminations if possible, becausethese reports and discriminations are mere behaviour and leave out what we want.

A critic of Farrell’s argument might object at this point. For, even grantingthat X’s report itself would be a piece of behaviour, it does not yet follow thatwhat X would tell us about would be mere behaviour. On the contrary, itseems that X might tell us about the phenomenal properties of the experiencethat he had when presented with a red shape. So we need to be provided with areason why X’s apparent description of an experience should not be taken atface value.

FARRELL ON BEHAVIOUR AND EXPERIENCE: MARTIANS AND ROBOTS 11

6 Farrell does not distinguish between epistemological and metaphysical ‘leaving out’claims.


A major theme in Farrell’s argument is the apparent contrast between‘behaviour’ and ‘experience’, in terms of which the problem is raised. Farrellpoints out that, in ordinary unproblematic cases where behaviour is con-trasted with experience, the term ‘behaviour’ is restricted to overt behaviour.But in the case of the putatively problematic contrast—where the sciences of the mind are supposed to leave out experience—the notion of behaviour is stretched to include ‘the covert verbal and other responses of the person,his response readinesses, all his relevant bodily states, and all the possible discriminations he can make’ (p. 177). Farrell insists that, once the notion ofbehaviour is extended in this way, we cannot simply assume that it continuesto contrast with experience rather than subsuming experience. This theme isdeveloped in discussion of two classic philosophical examples, Martians androbots.

1.3.1 Wondering what it is like: Martians, opium smokers,and bats

In the example of ‘the man from Mars’ (1950, p. 183), Farrell asks us to imag-ine that physiologists and psychologists have found out all they could find outabout a Martian’s sensory capacities and yet they still wonder what it would belike to be a Martian. He says that the remark, ‘I wonder what it would be like tobe a Martian’, seems to be sensible because it superficially resembles otherremarks, such as ‘I wonder what it would be like to be an opium smoker’ and ‘I wonder what it would be like to be, and hear like, a bat’ (ibid.).

If, in an ordinary unproblematic context, I wonder what it would be like tobe an opium smoker, then I may suppose or imagine that I take up smokingopium and that I thereby come to learn how the addiction develops, for example.What I would learn in the hypothetical circumstances of being an opiumsmoker might, Farrell says, outrun what could be learned by the ‘clumsy’ sci-entific methods available at a given time. But it would not be different inprinciple from what could be learned from third-person observation. Thus(pp. 172–3):

Quite often [a psychologist] places himself in the role of subject. … What is impor-tant to note is that by playing the role of observer-subject, he does not add anything tothe discoveries of psychological science that he could not in principle obtain from theobservation of X [another subject] alone.

According to Farrell, what I would learn about the experience of the opiumsmoker from the point of view of the observer-subject would not fall underthe term ‘behaviour’ in the sense restricted to overt behaviour, but it would fallunder the term in its extended sense that includes covert responses, responsereadinesses, discriminations, and so on.



In a similar way, I could unproblematically wonder what it would be like tobe a bat. I could suppose that a witch turns me into a bat and that, from theprivileged position of observer-subject, I learn something about the being’sdiscriminations and response readinesses. But, on Farrell’s view, if I were tospend a day or so as a bat then what I would learn would not outrun devel-oped bat physiology and psychology. And he is quite explicit that it wouldrequire no distinctively subjective concepts or conceptions (p. 173):

[N]o new concepts are required to deal with what [the psychologist’s] own subject-observation reveals which are not also required by what was, or can be, revealed by his[third-person] observation of [another subject].

In this unproblematic kind of wondering what it is like to be an opiumsmoker or a bat, what I would learn about would be covert responses andinternal discriminations, behaviour in the extended and inclusive sense of thatterm. This would also be the case if I unproblematically wondered what it would be like to be a Martian (p. 185): ‘the “experience” of the Martianwould … be assimilable under “behaviour”’.

The example of the Martian began, however, with a kind of wondering thatwas supposed to be quite different from this unproblematic wondering aboutbehaviour in the inclusive sense of the term. It was supposed to be a problem-atic wondering about something that would inevitably be left out by thesciences of the Martian mind—a wondering about experience as contrasted,not only with overt behaviour, but even with behaviour in the extended andinclusive sense of the term. Farrell’s point is that, while unproblematic won-dering is ‘sensible’, this putatively problematic wondering is ‘pointless’ (p. 185).We have no right to assume that this contrast—between experience andbehaviour in the inclusive sense—is legitimate.

A critic of Farrell’s argument might concede this point but also insist on another. We cannot simply assume that behaviour in the inclusive sensecontrasts with experience; but equally we cannot simply assume that it sub-sumes experience. Until we have a positive argument for subsumption, therelationship between behaviour and experience should remain an open ques-tion. We shall come to Farrell’s positive arguments shortly (section 1.4); but,before that, we review the second of the two classic philosophical examples,the robots.

1.3.2 Robots—and the criteria for having a sensationThe question under discussion in the example of the robot is whether we needto retain the contrast between behaviour and experience in order to say (1950,p. 189): ‘If a robot were to behave just like a person, it would still not have anysensations, or feelings.’

FARRELL ON BEHAVIOUR AND EXPERIENCE: MARTIANS AND ROBOTS 13


Farrell’s answer to the question comes in two stages. First, in ordinary talkabout robots, the unproblematic contrast between experience and overtbehaviour is adequate for the purpose. A robot, in the ordinary sense of theterm, duplicates the overt behaviour of a human being but not the covertresponses, bodily states, internal discriminations, and so on. So, second, if theexample of the robot is to present a problem for Farrell’s view then we must bethinking of a robot that duplicates, not only our overt behaviour, but all ourcovert responses and internal discriminations as well. But then, Farrell says, hehas already argued that we cannot presume upon a contrast between experi-ence and behaviour in this extended and inclusive sense.

In order to avoid the ‘muddle’ that results, according to Farrell, from this‘unobserved departure from the ordinary usage of “robot”’ (p. 190), we couldset aside that term for the time being. Then there are two ways that we mightdescribe a mechanical system that duplicates the overt and covert, externaland internal, behaviour of a person. On the one hand, we might allow, in linewith what Farrell regards as our ‘usual criterion’ for having a sensation, thatthe mechanical system has sensations. On the other hand, we might adopt amore demanding criterion for having a sensation and deny that the mechani-cal system has sensations on the grounds that it is not a living thing.

Does either way of describing the mechanical system present a problem forFarrell’s view about experience? If, on the one hand, we allow that a systemthat produces the right external and internal behaviour has experience thenclearly the example provides no reason to retain a contrast between experienceand behaviour in the inclusive sense. If, on the other hand, we insist that, whilemechanical systems produce behaviour, only a living thing has experiencethen, of course, we do retain a kind of contrast between experience and behav-iour in the inclusive sense. This more demanding criterion allows us to denyexperience to inanimate robots. But the contrast between mechanical systemsand living things has no relevance to questions about the mental lives ofhuman beings, Martians, or bats. Farrell thus concludes that the example of therobot does not present a problem for the behaviourist psychology of organisms.

Bringing his discussion of robots even closer to contemporary philosophyof consciousness, Farrell invites us to consider a series of imaginary examplesof robots that duplicate our external and internal behaviour and are increas-ingly like living things. He suggests that, as we progress along this series, it willbe increasingly natural to allow that the robots have experience—sensationsand feelings: (p. 191):

General agreement [to allow the attribution of experience] would perhaps beobtained when we reach a machine that exhibited the robot-like analogue of repro-duction, development and death.



Farrell’s position thus leaves no conceptual space for zombies. It licenses theattribution of experience to a hypothetical living thing that duplicates ourovert behaviour, covert responses, internal discriminations and bodily states.Consciousness is entailed a priori by life plus the right behaviour.

1.4 Experience from the third-person point of viewI have described Farrell’s view that experience is subsumed by behaviour andhave indicated some of the ways in which Farrell defended his position againstthe objection that we need the distinction between behaviour and experiencein order to say the things that we want to say about Martians and robots. But itwould be reasonable to ask what considerations motivated Farrell’s view in thefirst place.

Part of the answer is that Farrell regarded scientists’ concerns about con-sciousness as manifestations of their ‘occupational disease of traditional dualism’(p. 170)—the dualism against which Gilbert Ryle argued in The Concept of Mind(1949). The conscious mind, as conceived by the dualist, was supposed to fillwhat would otherwise be gaps in causal chains. It was supposed to provide themiddle part of a causal story that begins with physical processes leading fromstimulation of sensory surfaces and ends with physical processes leading to con-tractions of muscles. As against this dualism, Farrell argued that the causal storyleading all the way from sensory stimulation to overt behaviour could be told interms of factors that, aside from being covert and internal rather than overt andexternal, could be grouped with behaviour—causal factors such as covertresponses, discriminations, response readinesses, and bodily states.

There are also more specific points that figure in the motivation for Farrell’sview. I consider two: Farrell’s claim that experience is featureless and his rejec-tion of distinctive first-person knowledge of experience.

1.4.1 Featureless experienceImmediately after introducing the apparent contrast between behaviour andexperience, Farrell argues that experience, if it is contrasted with behaviour inthe extended and inclusive sense, is ‘featureless’ (1950, p. 178). We are to con-sider X in the role of observer-subject looking at a red patch and ask whetherthere is anything about X’s experience that he can discriminate. Farrell’sanswer is that there is not (ibid.):

If he does discriminate something that appears to be a feature of the experience, thissomething at once becomes, roughly, either a feature of the stimulus in the sort of waythat the saturation of the red in the red shape is a feature of the red shape, or a featureof his own responses to the shape. X merely provides us with further informationabout the behaviour that he does and can perform.

EXPERIENCE FROM THE THIRD-PERSON POINT OF VIEW 15


Here, Farrell presents two options for what we might be tempted to regard as adiscriminated feature of an experience. Either it becomes a feature of theworldly stimulus or else it becomes a feature of the subject’s response (that is,the subject’s behavioural response, in the inclusive sense of the term).

The first option is that a putative feature of experience is better conceived asa feature of the worldly stimulus. David Armstrong (1996) takes this as ananticipation of the representationalist proposal that the phenomenal proper-ties of an experience are determined by its representational properties—thatis, by how it represents the world as being. I shall consider representationalismlater (section 1.9). For now, let us note that a critic of Farrell’s argument mightask how the view that experiences have representational properties is sup-posed to be consistent with the claim that experiences are featureless. For,intuitively, how an experience represents the world as being is an aspect ofwhat it is like for a subject to undergo that experience, an aspect or feature ofits phenomenology.

A critic might also have a worry about the second option in the quoted pas-sage, the idea that the discriminated feature of an experience becomes afeature of the behavioural response. The critic might urge that it is not obvioushow the fact that X’s response is a piece of behaviour is supposed to support the claim that X’s response provides information only about behaviour.(In essence, this is the same objection that was entered at an earlier point inFarrell’s argument—see the beginning of section 1.3) We still need to be pro-vided with a reason why X’s behaviour should not be taken at face value, asevidence that he has discriminated a feature of his experience.

1.4.2 Acquaintance and the first-person point of viewFarrell himself anticipates an objection to his claim that experience is featureless,namely, that from the fact that experience has ‘no features that can be described,or discriminated, or reported in a laboratory’ it does not follow that experiencehas no features at all. He imagines an opponent saying (1950, p. 181):7

[Experience] may still possess features with which we can only be acquainted. … When,for example, we look at a red patch, we all just know what it is like to have the corre-sponding experience, and we all just know how it differs from the experience we havewhen looking at a green patch.


7 The imagined opponent’s proposal is a striking anticipation of McGinn’s comment(2004, p. 9): ‘if we know the essence of consciousness by means of acquaintance, then wecan just see that consciousness is not reducible to neural or functional processes (say)—just as acquaintance with the colour red could ground our knowledge that redness is notthe same as greenness, say’.


He also has the opponent propose that the problem lies in restricting observa-tion to the third-person case (p. 183). Farrell responds in his own person thatexperience remains featureless even if we allow first-person observation sinceapparent expressions of first-person knowledge about our experiences ofworldly objects are really based on our discrimination of our responses tothose objects (ibid.): ‘we are … liable to mistake features of our responses tothe [object] for some indescribable and ineffable property of the experience’.

At this stage, a critic might reckon Farrell’s response to be unsatisfying, sincethere is still no direct argument against the idea of features of experience thatcan be discriminated from a first-person point of view. But, after the discus-sion of the example of the man from Mars, Farrell returns to first-personknowledge (‘Knowing at first hand’, p. 185). Here, the opponent is imagined toobject that wondering what it is like to be a Martian is ‘wondering what itwould be like to have first-hand knowledge of the experience of a Martian’ andthat this first-hand knowledge would clearly be quite different from anythingthat one could learn by ‘hearing a description’. We already know that Farrell is bound to reject this objection by insisting that the observer-subject learnsabout covert responses and internal discriminations and that this knowledge isavailable, in principle, to third-person observation and conception. But he nowadvances a new response.

Knowledge at first hand, in the ordinary use of the term, is contrasted withknowledge at second hand, which is learning from someone else. But in thecase of knowing what it is like to be a Martian, Farrell’s opponent envisagesour knowing at first hand something that it is impossible to learn at secondhand, knowing by acquaintance something that it is impossible to learn bydescription. So, in the problematic case as it is conceived by the opponent,knowing at first hand ‘is not contrastable with anything [and so] this objectionsimply has not given a use to the expression “to know at first hand”’ (p. 186).

Here Farrell makes use of a contrast argument, a kind of argument that wasdeployed by Ryle in Dilemmas (1954). Ryle says, for example (1954, p. 94):‘There can be false coins only where there are coins made of the proper mate-rials by the proper authorities’; and (ibid., p. 95): ‘Ice could not be thin if icecould not be thick’. Similarly, Farrell is arguing that what could not be knownat second hand could not be known at first hand.

Contrast arguments can sometimes be persuasive. For example, if thin ice isdefined as ice that is thinner than average, then not all ice can be thin ice. Ifthere is to be ice that is thinner than average then there must also be some icethat is thicker than average. But, in general, contrast arguments do not succeedin showing that if an expression does not apply to anything then a contrastingexpression does not apply to anything either. A philosopher who claims that,

EXPERIENCE FROM THE THIRD-PERSON POINT OF VIEW 17


as a matter of necessity, there are no immaterial substances is not thereby sad-dled with the conclusion that there are no material substances either—nor withthe conclusion that the expression ‘material substance’ has not been given a use.

In the case of Farrell’s contrast argument, the expression:

(1) knows at first hand what it is like to be a Martian

contrasts with:

(2) knows at second hand what it is like to be a Martian.

The argument turns on the claim, made by Farrell’s opponent, that, as amatter of necessity, expression (2) does not apply to anyone. Nobody canknow at second hand what it is like to be a Martian. But—as is generally thecase with contrast arguments—Farrell’s argument does not succeed in show-ing that his opponent is saddled with the conclusion that expression (1)cannot apply to anyone either, nor with the conclusion that the opponent ‘hasnot given a use’ to expression (1).

1.5 Farrell, Dennett, and the critical agendaMore than 20 years before Nagel (1974), Farrell considered the question whatit is, or would be, like to be a bat. But, as we have now seen, Farrell used thequestion for purposes that were completely opposed to the ideas in Nagel’spaper. According to Farrell, the facts about experience do not outrun the factsthat are available to the sciences of the mind by third-person observation andthere are no distinctively subjective, first-person concepts that are deployed inour knowledge about experience. When physiologists and psychologists worrythat their accounts are incomplete because they leave out experience, ‘theirfears are groundless’ (1950, p. 197).

There are questions about experience that may seem to be problematic forFarrell’s behaviourist account of consciousness—questions about what it wouldbe like to be a bat or a Martian; about whether a robot could have experiences;about features of experience that a subject can discriminate; about acquain-tance with phenomenal properties; and about distinctively first-personknowledge. But Farrell argues that these apparently problematic questions rest on various philosophical errors—on the unwarranted assumption thatbehaviour, in the inclusive sense, continues to contrast with experience ratherthan subsuming it; on the failure to apply usual criteria; on the assumptionthat experience itself has features that can be discriminated; on the confusionbetween features of our responses to worldly objects and phenomenal proper-ties of experience; and on the failure to give meaning to the terms in whichquestions are cast.



Farrell’s view is strikingly similar to the account of consciousness thatDennett (1988, 1991, 2005) has developed in recent years, although there isalso a difference of dialectical context between them. The similarity is clear if we consider Farrell’s insistence that there is no knowledge available to the observer-subject that is not also available to third-person observation(section 1.3.1) alongside Dennett’s ‘A third-person approach to consciousness’(2005, chapter 2), or Farrell’s implied rejection of the conceivability of zom-bies (section 1.3.2) alongside Dennett’s ‘The zombic hunch: Extinction of anintuition?’ (2005, chapter 1), or Farrell’s rejection of indescribable and ineffa-ble properties of experience (section 1.4.2) alongside Dennett’s ‘Quiningqualia’ (1988).

The difference of dialectical context is this. Farrell was addressing a problemthat was raised by scientists—they feared that their accounts were bound toleave out experience. Farrell thought that philosophy could show that the scientists’ fears were groundless. In contrast, Dennett regards himself asremoving obstacles to progress towards a science of consciousness that havebeen erected, not by worried scientists, but by other philosophers—particularly,by philosophers who say that there is an explanatory gap.8

1.5.1 The need for a critical agendaThe first choice point in the philosophy of consciousness is whether to affirmor deny that there is an explanatory gap, that the physical sciences leave outconsciousness in the epistemological sense, that there is no a priori entailmentfrom physical truths to truths about conscious mental states conceived subjec-tively. Philosophers who deny that there is an explanatory gap (Dennett,Farrell) are able to proceed directly to type-A materialism. Those who allowthat there is an explanatory gap (Block, Chalmers, Levine, McGinn, Nagel)face a second choice: type-B materialism or dualism.

We observed earlier that a type-A materialist must deny that there are distinctively subjective conceptions of the kind that Nagel envisages.As Chalmers (2002) notes, a type-A materialist may appear as a reductionistor as an eliminativist about consciousness, promoting a behaviourist or func-tionalist conception of consciousness or saying that there is no such thing as

FARRELL, DENNETT, AND THE CRITICAL AGENDA 19

8 See the subtitle of his book, Sweet Dreams (2005), ‘Philosophical obstacles to a science ofconsciousness’, and the critical discussion of Block, Chalmers, Levine, McGinn, andNagel, therein. My own view, in contrast, is that it does not obstruct progress towards ascience of consciousness to point out that, if we have distinctively subjective conceptionsof types of experience, then truths about conscious mental states conceived subjectivelywill not be entailed a priori by physical truths.


consciousness as it is conceived subjectively (nothing in reality corresponds todistinctively subjective conceptions). As Chalmers also points out, type-Amaterialism involves ‘highly counterintuitive claims [that] need to be sup-ported by extremely strong arguments’ (2002, p. 251).

It is inevitable, then, that Farrell’s argument develops a partly criticalagenda. He rejects the very idea of distinctively first-person conceptions oftypes of experience; and he rejects the idea of a distinctive kind of knowledgegained by first-person acquaintance with the features of experience. On hisview, such conceptions, and the apparently problematic questions about expe-rience to which they give rise, are based on philosophical and conceptualerrors. The proper conceptions of conscious mental states or types of experiencecan be constructed out of objective conceptions of behaviour in the inclusivesense. There is a corresponding critical agenda in Dennett’s work. Just asFarrell argues that there are no discriminable features of experience withwhich subjects are acquainted, so Dennett argues that ‘there are no such prop-erties as qualia’ conceived as ‘directly and immediately apprehensible inconsciousness’ (1988, pp. 43, 47).

There remains, of course, a substantive question whether Farrell’s criticalagenda is effective, whether his negative arguments are sufficiently strong. At var-ious points, I have noted ways in which a critic might respond to his arguments.More generally, most contemporary philosophers of consciousness wouldreject Farrell’s apparent commitment to Rylean behaviourism and, particularly,his use of a contrast argument to cast doubt on the idea of knowing at firsthand what a type of experience is like. A similar question can, of course, beraised concerning the critical aspect of Dennett’s work.9


9 I noted earlier (section 1.1.2) that type-A materialism—the approach adopted byDennett—is probably the minority approach amongst philosophers of consciousnesswho defend physicalism. However, in this chapter I provide no details of Dennett’s position.A proper assessment of the critical aspect of his work would need to consider the third-person approach to studying consciousness that he calls ‘heterophenomenology’ (thephenomenology of another; see Dennett 1991). The connection between the heterophe-nomenological approach and the more explicitly critical aspect of Dennett’s work isapparent in the following passage in which Dennett criticises philosophers who assumethat, in addition to recognitional and discriminatory capacities, there is ‘a layer of “directacquaintance” with “phenomenal properties”’ (2007, p. 20): ‘These [recognitional/dis-criminatory] capacities are themselves the basis for the (illusory) belief that one’sexperience has “intrinsic phenomenal character,” and we first-persons have no privilegedaccess at all into the workings of these capacities. That, by the way, is why we shouldn’t doauto-phenomenology. It leads us into temptation: the temptation to take our own first-person convictions not as data but as undeniable truth.’


1.6 The knowledge argumentType-A materialism, the kind of position adopted by Farrell and Dennett, isboth conceptually and metaphysically reductionist. It is conceptually deflation-ist physicalism (Block 2002; Papineau 2002). The opposite position, dualism,is committed to both a duality of conceptions and a duality of properties.(A dualist may also be committed to a duality of states, processes, and eventsand perhaps—as in Descartes’s case—a duality of substances.) Type-B materi-alism is an intermediate position, combining conceptual dualism withmetaphysical reductionism. It is conceptually inflationist physicalism.

In recent philosophy of consciousness, Jackson’s (1982, 1986) knowledgeargument is one of two prominent attempts to argue from something like the explanatory gap or the epistemological ‘leaving out’ claim to the dualistconclusion that physicalism is false. The argument features Mary the brilliantscientist who, in her black-and-white room, learns everything about the physical world and then, on leaving the room for the first time, sees somethingred. The powerful intuition generated by the story of Mary is that, when she first sees a red rose or a ripe tomato she gains new knowledge. Now sheknows, whereas before she did not know, what it is like to see red. Since Maryalready knew all the physical facts, the knowledge argument concludes thatthere are facts that are not physical facts and that physicalism is thereforefalse.10

The other major argument for dualism in recent philosophy of conscious-ness is Chalmers’s (1996) conceivability argument, which also begins from anepistemological premise. This argument proceeds from the premise that zom-bies are conceivable (zombies are not a priori impossible) to the metaphysicalclaim that zombies are possible (zombies exist in some possible world) and,thence, to the conclusion that physicalism is false (the phenomenal propertiesof conscious mental states are not strongly determined or necessitated byphysical properties). The knowledge argument and the conceivability argu-ment raise many of the same issues—particularly concerning the transitionfrom epistemology to metaphysics—and they present philosophers of con-sciousness with the same options of type-A materialism, type-B materialism,and dualism. In the remainder of this chapter, the discussion of these optionsis organized around the knowledge argument.

THE KNOWLEDGE ARGUMENT 21

10 For a well-chosen sample of philosophical discussion of the knowledge argument, seeLudlow et al. (2004). The introduction by Stoljar and Nagasawa provides a helpfuloverview, as does the review of the book by Byrne (2006). The knowledge argument alsoplays a major role in David Lodge’s novel Thinks º (2001).


1.6.1 The structure of the knowledge argumentIt seems that, when Mary is released, she learns something new, somethingthat she could not have worked out a priori from what she knew before shesaw a red rose or a ripe tomato. This is an epistemological intuition that wouldsupport the claim that there is an explanatory gap—that is, the claim that thephysical sciences leave out consciousness in the epistemological sense. But it is not immediately clear how it could justify the conclusion that physicalism isfalse—that is, the conclusion that the physical sciences leave out consciousnessin the metaphysical sense.

Jackson makes the transition from epistemology to metaphysics by drawingon a crucial component of his overall philosophical position—a componentthat was not explicit in the earliest presentations (1982, 1986) of the knowl-edge argument. This is the claim that, if physicalism is true then there is an a priori entailment from the true physical story about the world to the truestory about conscious mental states and their phenomenal properties.11

Daniel Stoljar and Yujin Nagasawa explain this component of Jackson’s positionin terms of the psychophysical conditional, ‘If P then Q’, where P is the conjunc-tion of all the physical truths and Q is the conjunction of all the psychologicaltruths (2004, p. 15): ‘[W]e should assume that, in both the 1982 and 1986essays, Jackson was supposing… that: if physicalism is true, the psychophysicalconditional is a priori.’

In summary, we shall consider the knowledge argument as depending ontwo main premises. The first premise is epistemological:

(P1) Mary learns something new on her release.

The second premise is the principle linking epistemology and metaphysics:

(P2) If physicalism is true then the psychophysical conditional is a priori.

A powerful intuition supports the first premise (P1). On her release,Mary learns something that she could not have worked out a priori from what she knew in her black-and-white room, even though she already knew allthe physical truths. If the first premise is true then the psychophysical con-ditional is not a priori. In that case, by the second premise (P2), physicalism is false.


11 See Jackson (1995)—a postscript to Jackson (1986). For a more detailed account, seeJackson (1998a), Chapter 1, esp. pp. 6–14 and 24–7 on the entry by entailment thesis andChapter 3, esp. pp. 80–3 on the question of a priori deducibility.


1.6.2 An objection to the second premiseThe principle (P2) linking physicalism to a priori entailment certainly helpswith the knowledge argument’s transition from an epistemological first premiseto a metaphysical conclusion. But the principle itself seems to be open to theobjection that the metaphysical determination or necessitation of phenome-nal facts by physical facts could be a posteriori rather than a priori. In SaulKripke’s (1980) famous example, it is a necessary a posteriori truth that wateris H2O. So the fact that H2O covers most of the planet determines a posteriorithat water covers most of the planet. Why cannot the a posteriori determina-tion of facts about water by facts about H2O serve as a model for thedetermination of phenomenal facts by physical facts?

Jackson responds to this objection by arguing that familiar examples ofa posteriori determination, such as the example of water and H2O, do not support the idea of a posteriori determination of phenomenal facts by thetotality of physical facts.12 The fact that water covers most of the planet isdetermined a posteriori by the fact that H2O covers most of the planet; but it isdetermined a priori by a richer set of facts about H2O. The reason is that, onJackson’s view, it is a priori—indeed, a matter of conceptual analysis—thatwater is whatever stuff is colourless, odourless, falls from the sky, and so on. Itis a priori that water is whatever stuff ‘fills the water role’.13 Consequently,there is an a priori entailment from the facts that H2O covers most of theplanet and that H2O fills the water role to the fact that water covers most ofthe planet.

In a similar way, Jackson says (1995/2004, p. 414):

A partial story about the physical way the world is might logically necessitate the psy-chological way the world is without enabling an a priori deduction of thepsychological way the world is. … But the materialist is committed to a complete ornear enough complete story about the physical way the world is enabling in principlethe a priori deduction of the psychological way the world is. … I think it is crucial forthe truth of materialism (materialism proper, not some covert form of dual attributetheory of mind) that knowing a rich enough story about the physical nature of ourworld is tantamount to knowing the psychological story about our world.


12 See Jackson (1995, 2003), Braddon-Mitchell and Jackson (2007, pp. 139–40).

13 This part of Jackson’s view would be roughly captured by saying that the term ‘water’ is adescriptive name with its reference fixed by the description ‘the stuff that is colourless,odourless, falls from the sky, and so on’ or by saying that the term ‘water’ behaves seman-tically and modally like the description ‘the stuff that actually (that is, in the actual world)is colourless, odourless, falls from the sky, and so on’. See Davies and Humberstone (1980)for an early development of this view in the framework of two-dimensional semantics.


We shall return later (sections 1.8.2 and 1.8.3) to the issue of distinguishingphysicalism from ‘some covert form of dual attribute theory’. For the moment,I want to continue with the question how the second premise of the knowl-edge argument—the principle linking physicalism to a priori entailment—isto be motivated.

1.6.3 The second premise and type-A materialismThe second premise says, in effect, that if physicalism is true then type-A mate-rialism is true. So long as both type-A and type-B materialism are options forthe physicalist, the premise is open to the obvious objection that physicalismmight be true without type-A materialism being true, because type-B material-ism might be true. Now, according to type-B materialism, the determinationof phenomenal facts by physical facts is a posteriori rather than a priori. So it iscertainly relevant to point out, as Jackson does with the example of water andH2O, that a posteriori determination by a partial set of facts may be consistentwith a priori determination by a richer set of facts. But, although this is relevant,it does not yet go to the heart of the matter.

Both a type-A materialist and a type-B materialist will agree that, if our con-ception of water is a physical-functional conception, then water facts areentailed a priori by H2O facts. This is not to say that all type-B materialistsaccept that we do have a physical-functional conception of water. For example,Brian McLaughlin says (2005, p. 280, n. 31): ‘I regard that as an unresolvedissue.’ But it is certainly open to a type-B materialist to agree with Jackson thatthe state of water covering most of the planet can be explained in terms offacts about H2O via functional analysis of the concept of water.

The type-B materialist disagrees with the type-A materialist, however, overthe question whether water facts are relevantly similar to phenomenal factsand, particularly, whether a functional conception of water is a good modelfor our conceptions of types of experience. Nagel’s account of the contrastbetween subjective conceptions and objective conceptions, and Levine’s claimthat there is an explanatory gap, both depend on our subjective conceptions oftypes of experience not being functional conceptions. Indeed, McLaughlinsays (2005, p. 280): ‘On one interpretation, [Levine’s] explanatory gap thesis isthe thesis that states of phenomenal consciousness cannot be physicallyexplained via º functional analysis.’

A type-A materialist denies that there is an explanatory gap and denies thatthere are distinctively subjective conceptions of types of experience. Type-Amaterialism is conceptually deflationist physicalism. So a defence of the secondpremise of the knowledge argument against the objection that physicalism mightbe true without type-A materialism being true must go beyond the uncontested



example of water and H2O. It must include an argument against the conceptuallyinflationist option of type-B materialism—an argument to show that subjectiveconceptions and the explanatory gap are inconsistent with physicalism.

1.6.4 The simplified knowledge argumentThe second premise, (P2), of the knowledge argument—made explicit by Stoljarand Nagasawa (2004, p. 15)—appears to be motivated by the backgroundassumption that conceptually deflationist physicalism (type-A materialism) isthe only physicalism worthy of the name, so that type-B materialism can berejected. The prospects for type-B materialism will be assessed later (sections1.11 and 1.12). In the meantime, we can consider a simplified version of theknowledge argument from the epistemological first premise (as before):

(P1) Mary learns something new on her release.

and a second premise that is now true by definition:

(P2A) If type-A materialism is true then the psychophysical conditional is a priori.

to the conclusion that type-A materialism is false.

The simplified knowledge argument seems to be valid and the second premise(P2A) is true by the definition of type-A materialism. The argument presentsus with three options. If we reject the conclusion and accept type-A material-ism then we must also reject the epistemological premise (P1). If we accept theepistemological premise then we must also accept the conclusion, reject type-A materialism, and choose between type-B materialism (conceptuallyinflationist physicalism) and dualism.14


14 The claim that the simplified knowledge argument presents just three options—type-Amaterialism, type-B materialism, and dualism—involves a degree of simplification.I assume that it is legitimate to include in the setting-up of the example that Mary alreadyknows all the physical facts while she is in her black-and-white room. I also assume that, ifMary learns something new on her release, then she gains propositional knowledge. Eachof these assumptions might be rejected, providing two more options (see Byrne 2006).According to the first option, there are physical facts of which the physical sciences tell usnothing (Stoljar 2001, 2006). According to the second option, Mary gains ‘know how’,rather than propositional knowledge, on her release. This is the ability hypothesis, dis-cussed in section 1.10.2. Finally, the validity of the simplified knowledge argument mightbe challenged. Someone might deny that the first premise (P1) really entails that the psy-chophysical conditional is not a priori. As we observed earlier (section 1.2.3), the claimthat there is no a priori entailment from physical truths to truths about conscious mentalstates conceived subjectively is not an immediate consequence of the claim that subjectiveconceptions cannot be constructed from objective conceptions. For discussion of this finaloption, see Byrne (2006), Nida-Rümelin (1995, 2002) and Stoljar (2005).


The simplified knowledge argument no longer depends on the assumptionthat type-B materialism can be rejected. The role of that assumption, if itcould be justified, would be to license the transition from the limited conclu-sion of the simplified knowledge argument to the more sweeping conclusionof the original knowledge argument, namely, that physicalism is false.

1.7 The argument for physicalismThere is much more to be said about Jackson’s knowledge argument againstphysicalism. But the first thing to be said is that Jackson himself has come toreject the knowledge argument. He is now convinced that physicalism must betrue (2004, p. xvi):

On the face of it, physicalism about the mind across the board cannot be right. [But] I now think that what is, on the face of it, true is, on reflection, false. I now think thatwe have no choice but to embrace some version or other of physicalism.

1.7.1 The causal argument for physicalismDavid Papineau summarizes the causal argument for physicalism—which hedescribes as ‘the canonical argument’—as follows (2002, p. 17):

Many effects that we attribute to conscious causes have full physical causes. But itwould be absurd to suppose that these effects are caused twice over. So the consciouscauses must be identical to some part of those physical causes.

Following Papineau, we can set out the causal argument a little more formally.There are three premises, of which the second is ‘the completeness of

physics’ (2002, pp. 17–18):

(1) Conscious mental occurrences have physical effects.(2) All physical effects are fully caused by purely physical prior histories.(3) The physical effects of conscious causes aren’t always overdetermined by distinct

causes.

From these premises, Papineau says, materialism follows (ibid., p. 18)—wherematerialism is the thesis that conscious states are either (a) identical withphysical states (in the strict sense of states of kinds studied by the physical sciences) or else (b) identical with ‘“physically realized functional states”,or with some other kind of physically realized but not strictly physical states’(p. 15). (Papineau uses the term ‘physicalism’ for the stricter thesis that thefirst disjunct (a) is true. In previous sections, I have not distinguished betweenphysicalism and materialism.)

It is possible to evade the causal argument by rejecting the completeness of physics (denying premise 2) and allowing, instead, that some physicaloccurrences have irreducibly non-physical causes. One historical view of this



kind proposed the operation of vital forces or special powers of living matter.Papineau (2002, Appendix) explains in some detail how developments in bio-chemistry and neurophysiology during the first half of the twentieth century‘made it difficult to go on maintaining that special forces operate inside livingbodies’ (2002, pp. 253–4). Nevertheless, Chalmers suggests that, if we were tohave independent reason to reject physicalism, then we should leave open thepossibility of rejecting the completeness of physics and maintaining dualistinteractionism instead—‘holding that there are causal gaps in microphysicaldynamics that are filled by a causal role for distinct phenomenal properties’(2002, p. 261).

An alternative way to evade the causal argument is to accept the complete-ness of physics, so that all physical effects have full physical causes, but then toavoid the unwanted consequence that these effects are ‘caused twice over’ bydenying that conscious mental occurrences have physical effects (denyingpremise 1). This is epiphenomenalism about conscious mental states, the posi-tion that Jackson (1982) adopted when he accepted the knowledge argumentagainst physicalism.

1.7.2 Strict and relaxed versions of physicalism: identity orsupervenience

Even if we accept all the premises of the causal argument as Papineau presentsit, we can still evade the conclusion that conscious states are identical withphysical states in the strict sense (physicalism, in Papineau’s terminology). Tosee this, we need to make some distinctions within the idea of an effect beingcaused twice over—that is, refine the idea of overdetermination by ‘distinctcauses’ (refining premise 3).

A man’s death is overdetermined by distinct causes, or caused twice over,if he is ‘simultaneously shot and struck by lightning’ (2002, p. 18). That kindof causation by two independent causes is, we can agree, an unintuitive modelfor causation by conscious mental states. But we should also consider the case of causes that are numerically distinct but not independent. In particular,we should consider supervenient properties—that is, higher-level propertieswhose instantiation is strongly determined or necessitated by the instantiationof certain lower-level properties. As Papineau observes (ibid., p. 32), it is quitenatural to regard these higher-level properties as having causal powers in virtue of the causal powers of the lower-level properties on which theysupervene. In short, supervenient properties may have supervenient causalpowers.

Suppose we allow that some properties that supervene on physical proper-ties might not, strictly speaking, be physical properties themselves. Then we

THE ARGUMENT FOR PHYSICALISM 27


make room for the possibility that conscious causes might be numerically dis-tinct from physical causes, yet without any suggestion that there would be twoindependent causes of the same physical effect. The causal powers of the phe-nomenal properties of conscious mental states would supervene on—wouldbe determined or necessitated by (perhaps even constituted by)—the causalpowers of lower-level physical properties. So it would only be in a very attenu-ated sense that the effects of the conscious causes would be caused twice over.It would be quite different from the case of a man being simultaneously shotand struck by lightning. It would be more like the case of a man being bruisedby simultaneously bumping into both a bronze statue and the lump of bronzethat constitutes the statue.15

Thus, in the end, the causal argument allows for a strict version of physical-ism about the mind, according to which all mental properties are physicalproperties (defined as properties that figure in the physical sciences), and alsofor a more relaxed version, according to which all mental properties at leastsupervene on physical properties. On both identity (strict) and supervenience(relaxed) versions of physicalism, mental properties have causal powers (theyare not epiphenomenal) but there is no evident threat of overdeterminationby distinct and independent causes.

1.8 Jackson’s rejection of the knowledge argumentThe causal argument for physicalism allows for a relaxed version ofphysicalism—supervenience physicalism—and, as Stoljar and Nagasawa note(2004, p. 14): ‘in contemporary philosophy, physicalism is usually construed interms of what is called a supervenience thesis’. It might be tempting to assume,therefore, that the causal argument for physicalism adequately captures Jackson’sreason for rejecting the knowledge argument against physicalism. However, inthis section, I shall explain how Jackson’s own grounds for rejecting the knowl-edge argument go beyond the causal argument for physicalism and how his ownconception of physicalism is more demanding than supervenience physicalism.

1.8.1 Knowledge and epiphenomenalismWhen he put forward the knowledge argument against physicalism, Jacksonalready accepted that physical effects have full physical causes (the completeness


15 On some philosophical views (including mine), although the statue and the lump are inthe same place at the same time, they are strictly speaking different objects with some dif-ferent properties. Nevertheless, the man’s bruise is no worse for his having bumped intoboth of them.


of physics) and that instantiations of non-physical properties have no causal con-sequences in the physical order (dualist interactionism is false). As a consequence,he accepted that phenomenal properties of experience, if they are not physicalproperties, are epiphenomenal. Against such a view, the causal argument forphysicalism makes no headway (because premise 1 is not accepted). From themid-1990s, Jackson came to argue (1998b, 2005a; Braddon-Mitchell and Jackson1996) that, since the denial of physicalism involves epiphenomenalism aboutqualia, the knowledge argument is undermined by its own conclusion.

Jackson says that his reason for changing his mind about the knowledgeargument is that (1998b/2004, p. 418): ‘Our knowledge of the sensory side ofpsychology has a causal source.’ When Mary emerges from her black-and-white room and sees something red, she undergoes a change—from notknowing what it is like to see red to knowing what it is like to see red. This is,or involves, a physical change. The physical change is caused by something and,by the completeness of physics, it has a full physical cause. If the phenomenalproperties of Mary’s experience of seeing a red rose or a ripe tomato are non-physical, and so epiphenomenal, then those properties of Mary’s experiencecan play no part in the causation of Mary’s coming to know what it is like to seered. Thus (Braddon-Mitchell and Jackson 1996, p. 134): ‘Mary’s discovery … ofsomething important and new about what things are like is in no sense due tothe properties, the qualia, whose alleged instantiation constituted the inade-quacy of her previous picture of the world.’

As Jackson came to see the situation, the conclusion of the knowledge argu-ment has the consequence that phenomenal properties are epiphenomenal,and this undermines the intuition that Mary gains new knowledge on herrelease as a result of experiencing for herself what it is like to see red.16 Thiswas enough to persuade Jackson that ‘there must be a reply’ to the knowledgeargument (Braddon-Mitchell and Jackson 1996, p. 143), ‘it must go wrong’(Jackson 2005a, p. 316).

1.8.2 Jackson’s version of physicalismPhysicalism as Jackson conceives it is not the strict version. It is not committedto the claim that phenomenal properties are identical to properties that figure

JACKSON’S REJECTION OF THE KNOWLEDGE ARGUMENT 29

16 This objection against the knowledge argument was raised by Michael Watkins (1989):‘if Jackson’s [1982] epiphenomenalism is correct, then we cannot even know about ourown qualitative experiences’ (p. 158); ‘Jackson’s epiphenomenalism provides us with noavenues by which we might justifiably believe that there are qualia. If epiphenomenalism iscorrect, then Mary, the heroine of Jackson’s knowledge argument against physicalism,gains no new knowledge when she leaves her black and white room’ (p. 160).


in the present or future science of physics, nor even to the claim that phenom-enal properties are identical to properties that figure in the physical sciencesconceived more broadly, to include physics, chemistry, biology, and neuro-science. The reason is that physicalism is ‘a theory of everything in space-time’(2006, p. 231) and ‘the patterns that economics, architecture, politics and veryarguably psychology, pick out and theorize in terms of, include many that donot figure in the physical sciences’ (ibid., p. 234). The properties in terms ofwhich Jackson’s physicalism is defined are not just physical properties ‘in thecore sense’ (properties that figure in the physical sciences) but also includephysical properties ‘in an extended sense’ (p. 233).

Jackson’s version of physicalism is not the relaxed version either. The reasonis this (2006, p. 243): ‘A live position for dual attribute theorists is that psycho-logical properties, while being quite distinct from physical properties, arenecessitated by them.’ So, supervenience physicalism characterized withoutsome additional requirement is not properly distinguished from ‘a necessitar-ian dual attribute view’ (ibid.). According to Jackson, if superveniencephysicalism is not to be ‘a dual attribute theory in sheep’s clothing’ (p. 227)then the determination of supervening properties by core physical propertiesmust be necessary and a priori.

Thus, Jackson proposes that physical properties in the extended sense areproperties whose distribution is determined a priori by the distribution ofphysical properties in the core sense. Two simple examples may help to makethis idea clearer. First, while the property of being silver and the property ofbeing copper both figure in the science of chemistry, it is not clear that chem-istry or any other physical science has a use for the disjunctive property ofbeing either silver or copper. So the disjunctive property might not be a physi-cal property in the core sense. But it is a physical property in the extendedsense because whether something instantiates the disjunctive property isdetermined a priori by whether it is silver and whether it is copper. Second,while jewellers talk about sterling silver it is not clear that there is a science ofthings that are made up of 92.5% silver and 7.5% copper. So the property ofbeing sterling silver might not be a physical property in the core sense. But it isa physical property in the extended sense because the distribution of sterlingsilver is determined a priori by the distributions of silver and of copper.

1.8.3 The case for a priori physicalismJackson’s version of physicalism is a priori physicalism and, in fact, the notionof the a priori enters twice over. First, a priori physicalism requires that allproperties should be physical properties, defined as properties that are deter-mined a priori by properties that figure in the physical sciences (section 1.8.2).



Second, a priori physicalism requires that all the facts, particularly the psycho-logical facts, should be entailed a priori by the physical facts (section 1.6.1).The two requirements are not obviously equivalent. If there were subjectiveconceptions of physical properties then the requirement of a priori entailmentwould not be met (there would be an explanatory gap) although the require-ment of a priori determination of properties could still be met. In both cases,however, the a priori element in the account is promoted as distinguishingphysicalism from ‘a dual attribute theory in sheep’s clothing’ (2006, p. 227) orfrom ‘some covert form of dual attribute theory of mind’ (1995/2004, p. 414;see above, section 1.6.2).17

Jackson places his a priori physicalism in the tradition of Australian materi-alism—the materialism of J.J.C. Smart (1959) and David Armstrong(1968)—according to which ‘spooky properties are rejected along with spookysubstances’ (2006, p. 227). He is opposed to all dual attribute theories, includ-ing even the ‘necessitarian’ dual attribute theory that says that phenomenalproperties are distinct from, but strongly determined (necessitated) by, physi-cal properties. The problem with dual attribute theories, Jackson says, is that‘spooky properties… would be epiphenomenal and so both idle and beyondour ken’ (ibid.). As we saw (section 1.8.1), it was because of this problem thatJackson rejected the knowledge argument and its conclusion that the phe-nomenal properties of experience are ‘spooky’, non-physical, properties.

It may be, however, that non-physical properties need not be epiphenomenal.As Terence Horgan puts the point (1984/2004, p. 308, n. 6): ‘Indeed, even ifqualia are nonphysical they may not be epiphenomenal. As long as they aresupervenient upon physical properties, I think it can plausibly be argued thatthey inherit the causal efficacy of the properties upon which they supervene.’(In section 1.7.1, we noted that Papineau (2002, p. 32) makes a similar pro-posal.) The possibility of non-physical, but causally potent, properties raisestwo potential worries about Jackson’s position. First, it allows a response toJackson’s specific reason for rejecting the knowledge argument. Second, itraises the question whether there is any good objection to dual attribute theo-ries of the necessitarian variety.

JACKSON’S REJECTION OF THE KNOWLEDGE ARGUMENT 31

17 Jackson says that the thesis about the a priori determination of physical properties is a priori physicalism ‘understood as a doctrine in metaphysics, understood de re if you like’(2006, p. 229). This thesis is already sufficient to distinguish a priori physicalism from anecessitarian dual attribute theory. The thesis about a priori entailment is a priori physi-calism understood de dicto. Although a priori physicalism understood de dicto seems to gobeyond a priori physicalism understood de re, Jackson (2005b, p. 260) advances an argu-ment ‘that takes us from the de re thesis to the de dicto thesis’.


Jackson could respond to these worries by maintaining that his a prioriphysicalism is a better, because more austere, theory than any dual attribute view.He characterizes a priori physicalism as ‘bare’ physicalism in a passage thatmanifests something of W. V. O. Quine’s (1953; see Jackson 2005b, p. 257)‘taste for desert landscapes’ (2003/2004, pp. 425–6):

The bare physicalism hypothesis … that the world is exactly as is required to make thephysical account of it true in each and every detail but nothing more is true of thisworld in the sense that nothing that fails to follow a priori from the physical account istrue of it … is not ad hoc and has all the explanatory power and simplicity we can rea-sonably demand.

1.9 Physicalism and representationalismJackson now rejects the conclusion of the knowledge argument. As mentionedearlier, he thinks that ‘we have no choice but to embrace some version or otherof physicalism’ (2004, p. xvi). The specific version of physicalism that heaccepts is type-A materialism—also known as a priori physicalism, conceptu-ally deflationist physicalism, or bare physicalism. Consequently, he rejects theepistemological first premise of the knowledge argument. This, he now says, iswhere the argument goes wrong (2004, p. xvii–xviii): ‘[Mary] learns nothingabout what her and our world is like that is not available to her in principlewhile in the black and white room.’

This is what Farrell would say and Dennett does say.18 It is what Jacksonneeds to say; but it is not easy to defend. It certainly contrasts sharply with whathe said when he first put forward the knowledge argument (1986/2004, p. 52):‘[I]t is very hard to believe that [Mary’s] lack of knowledge could be remediedmerely by her explicitly following through enough logical consequences of hervast physical knowledge.’ In defence of his new position, Jackson needs tomake it plausible that, when Mary ‘knows what it is like to see red’, what shereally knows is something that is entailed a priori by the totality of physicalfacts about the world, facts that she already knew in her room.

1.9.1 RepresentationalismI mentioned earlier (section 1.3.1) that Armstrong interprets Farrell’s claimabout experience being featureless as an anticipation of the representationalistview that the phenomenal properties of an experience are determined by itsrepresentational properties—that is, by how it represents the world as being.Similarly, Stoljar and Nagasawa (2004, p. 25, n. 11) see Farrell’s idea of featureless


18 See Dennett (1991, pp. 398–406, reprinted in Ludlow et al. 2004, pp. 59–68; 2005, Chapter 5,‘What RoboMary knows’).


experience as related to the doctrine of the transparency or diaphanousness ofexperience, a doctrine that several contemporary philosophers take to stand ina close relationship to representationalism.19 In any case, it is to representa-tionalism that Jackson turns for his account of what Mary really knows whenshe knows what it is like to see red (2003/2004, p. 430): ‘we have to understandthe qualities of experience in terms of intensional [representational] properties’.

The starting point for representationalism is that ‘experience is essentiallyrepresentational … it is impossible to have a perceptual experience withoutthereby being in a state that represents that things are thus and so in the world’(Jackson 2007, p. 57). As a claim about perceptual experience, such as Mary’sexperience of a red rose or a ripe tomato, this is highly plausible. It is of thenature of perception to represent how things are in the world. The claim thatall experiences, including bodily sensations, are representational is less com-pelling but even strong opponents of representationalism may be prepared togrant it. Thus, for example, Ned Block says, ‘I think that sensations—almostalways—perhaps even always—have representational content’ (2003, p. 165).

Going beyond this starting point, representationalism is usually formulatedas a supervenience thesis. As between two conscious mental states, such as twoperceptual experiences, there can be no difference in phenomenal propertieswithout a difference in representational properties. The phenomenal characterof a conscious mental state is determined by its representational content(Byrne 2001). Representationalism is an unclear thesis to the extent that thenotion of representation itself is not well specified. It is also a controversialthesis. But the attraction of representationalism for a type-A materialist is thatit promises physical-functional conceptions of types of experience.

In philosophy of mind over the last quarter-century or so, the topics of con-sciousness and representation have mainly been considered somewhatseparately. As a result, even those who think that consciousness defies scien-tific explanation are apt to be confident that representation can be analysed in‘naturalistic’, physical-functional, terms. It is against this background thatJackson says (2003/2004, p. 432):

The project of finding an analysis of representation is not an easy one—to put itmildly. But … the answers that have been, or are likely to be, canvassed are all answers

PHYSICALISM AND REPRESENTATIONALISM 33

19 It is difficult to spell out a compelling argument from the transparency or diaphanous-ness of experience to representationalism (Stoljar 2004; see also Burge 2003, pp. 405–7).Tye (2000, p. 45) says: ‘I believe that experience is transparent. I also believe that its trans-parency is a very powerful motivation for the representationalist view. I concede, however,that the appeal to transparency has not been well understood.’ Jackson (2007, p. 57) says:‘I conclude that the famous diaphanousness or transparency of experience is not per sethe basis of an argument for representationalism.’


that would allow the fact of representation to follow a priori from the physicalaccount of what our world is like.

1.9.2 Representationalism and phenomenologyLet us agree, for the sake of the argument, that the representational facts areentailed a priori by the physical facts, just as the water facts are entailed a prioriby the H2O facts. This is not yet sufficient for a physicalist account of whatMary really knows when she ‘knows what it is like to see red’. For what Maryknows entails that there is something that it is like to see red; seeing red is a con-scious mental state. But, even according to representationalism, it is not thecase that the representational content of an experience ‘suffices to make anystate that has it conscious’ (Seager and Bourget 2007, p. 263). Or, as Alex Byrneputs it (2001, p. 234): ‘Intentionalism [representationalism] isn’t much of atheory of consciousness.’

Representationalism says that the phenomenal character of a consciousmental state is determined by its representational content. But it does not saythat the representational properties of a mental state determine that it is a con-scious mental state. There can be representation without consciousness. Sosomething needs to be added to representationalism if it is to provide anaccount of what Mary knows about the experience of seeing red. Jackson him-self says that the nature of an experience, including the fact that it is aconscious mental state, is determined by ‘the [representational] content of[the] experience plus the fact that the experience represents the content asobtaining in the way distinctive of perceptual representation’ (2007, p. 58; alsosee 2005a, p. 323).

He goes on to list five features that are putatively distinctive of perceptualrepresentation. The content of perceptual representation is rich, and inextri-cably rich; the representation is immediate; there is a causal element in itscontent; and perceptual experience plays a distinctive functional role inrespect of belief. If a state has representational content with these five featuresthen, Jackson says, ‘we get the phenomenology for free’ (2003/2004, p. 438).What is most important about these five features is that they can, let us sup-pose, be explicated in physical-functional terms.

The story of Mary generates the intuition that, on her release, Mary learnssomething new about the experiences of people who looked at red roses andripe tomatoes while she was in her room. As required by type-A materialism,Jackson rejects this intuition. He says that what Mary really knows is that the people were in physical states with a particular representational property(roughly, representing something as being red) and meeting five further conditions. Since both representation and the further conditions can be



explicated in physical-functional terms, this knowledge is not new but was, inprinciple, already available to Mary while she was in her room.

1.10 The epistemological intuition and the abilityhypothesis

Type-A materialism is counter-intuitive. Accepting it commits Jackson torejecting the epistemological intuition that Mary learns something new on herrelease. He needs to develop a critical agenda supporting that rejection. LikeFarrell and Dennett, Jackson needs to undermine the idea that there is more toknow about human experience than is entailed a priori by the totality of phys-ical facts.

1.10.1 Representationalism and the epistemological intuitionJackson stresses that representationalism highlights the distinction between arepresentational property and an instantiated property (2003, 2005a, 2007).Representationalism thus provides a reason to say that ‘there is no such property’(2003/2004, p. 430) as the ‘redness’ of the experience of seeing a rose. Rednessis not a property that experiences of roses instantiate; it is the property thatexperiences represent roses as instantiating.

Experiences do, of course, instantiate the property of representing things asbeing red. But this representational property of experiences is not a new prop-erty that was unknown to Mary before her release. It is a physical-functionalproperty that Mary knew about (or could have known about) in her black-and-white room. What is new after her release is that Mary now has an experienceinstantiating this property (and meeting five further conditions). We musttake care not to mistake a new instantiation of a representational, and there-fore physical, property of experiences for the instantiation of a new, andtherefore non-physical, property of experiences. A new instantiation of aproperty is not the instantiation of a new property.

This is an important point, but it is not clear that it undermines the intu-ition that supports the epistemological first premise of the knowledgeargument (Alter 2007). Jackson says (2007, p. 61): ‘The challenge from theknowledge argument is the intuition that the “red” of seeing red is a new sort ofproperty.’ But, on the face of it, the intuition that Jackson needs to undermine—the intuition that drives the knowledge argument—is not an explicitlymetaphysical intuition that, on her release, Mary learns about a new property ofexperiences. It is the epistemological intuition that Mary gains new knowledge—that she comes to know a fact that is not entailed a priori by the totality ofphysical facts that she already knew in her room.

THE EPISTEMOLOGICAL INTUITION AND THE ABILITY HYPOTHESIS 35


It is, of course, part of Jackson’s overall position that new knowledge wouldhave to be knowledge about new properties. That is what the second premise ofthe original knowledge argument says: if physicalism is true (no new properties)then the psychophysical condition is a priori (no new knowledge). But thatconnection between epistemology and metaphysics is not provided by repre-sentationalism about perceptual experiences. It depends on the assumptionthat type-B materialism can be rejected (section 1.6.3).

Representationalism may be developed in the service of physicalism aboutconscious mental states. It certainly plays a major role in contemporary phi-losophy of consciousness. But representationalism does not favour Jackson’sconceptually deflationist physicalism (type-A materialism) over conceptuallyinflationist physicalism (type-B materialism). Tye (1995, 2000) develops a ver-sion of representationalism that is very similar to Jackson’s.20 Yet Tye maintainsthat, on her release, Mary gains new subjective conceptions of physical—specifically, representational—properties, deploys those conceptions inpropositional thinking, and achieves new propositional knowledge.

1.10.2 The ability hypothesisWe have just seen that representationalism does not provide any independentmotivation for rejecting the epistemological intuition that Mary learns some-thing new on her release. But Jackson’s physicalist account of consciousexperience goes beyond representationalism.

According to representationalism, the properties of Mary’s experience whenshe sees a red rose for the first time are physical properties. Specifically, theyare properties of having such-and-such representational content and meetingfurther physical-functional conditions. They are not new properties but properties that Mary was already in a position to know about in her black-and-white room. What is new is that Mary now has an experience that instantiatesthose properties. Jackson describes Mary’s situation as follows (2003/2004,p. 439):

[S]he is in a new kind of representational state, different from those she was in before.And what is it to know what it is like to be in that kind of state? Presumably, it is to beable to recognize, remember, and imagine the state. … We have ended up agreeingwith Laurence Nemirow and David Lewis on what happens to Mary on her release.


20 According to Tye’s PANIC theory, phenomenal properties are determined by (indeed, areidentical with) properties of having Intentional Content that meets three conditions: it isPoised (poised to have a direct impact on beliefs and desires), Abstract (does not involveparticular objects), and Nonconceptual (in order for a state to have this kind of content itis not necessary for the subject of the state to be able to conceptualise the content). Thus,in Tye’s account, P+A+N plays the role that the five features play in Jackson’s account.


Here, Jackson goes beyond the basic claim of his response to the knowledgeargument, namely, that Mary does not gain new propositional knowledge onher release. He concedes something to the epistemological intuition. He saysthat Mary does gain something new and he allows that this might be describedas new ‘knowing what it is like’. But following the ability hypothesis response tothe knowledge argument proposed by Lewis (1988) and Nemirow (1980,2007), he says that what Mary gains is not new propositional knowledge butonly new abilities.

Papineau comments on the ability hypothesis (2002, pp. 59–60):

Some philosophers are happy to accept that Mary acquires new powers of imaginativere-creation and introspective classification, yet deny that it is appropriate to view thisas a matter of her acquiring any new phenomenal concepts. These are the sophisti-cated deflationists of the ‘ability hypothesis’.

Jackson proposes to join these ‘sophisticated’ conceptually deflationist physi-calists and, as Papineau says, their position—Mary gains new ‘know how’ ornew abilities—certainly seems preferable to ‘outright denial’ that Mary comesto know anything new at all. Subtle type-A materialism seems preferable tothe more straightforward type-A materialism of Farrell and Dennett. But,Papineau continues (ibid., p. 61):

Even so, the ability hypothesis does not really do justice to the change in Mary. If welook closely at Mary’s new abilities, we will see that they are inseparable from herpower to think certain new kinds of thoughts.

The abilities to which Jackson appeals in his description of Mary’s new‘knowing what it is like’ are the same abilities to which Tye (2000) appeals indescribing a subject’s possession of a subjective conception of a relativelycoarse-grained type of experience, such as seeing red (section 1.2.2). Theadvocate of the ability hypothesis needs to say why having these abilities is notsufficient for possession of a subjective conception that can be deployed inpropositional thinking about a type of experience.

In earlier writings, Jackson himself resists the ability hypothesis response tothe knowledge argument. He argues that Mary’s new abilities (to remember,imagine, recognize) are associated with a new cognitive capacity to engage innew propositional thinking and that Mary acquires ‘factual knowledge aboutthe experiences of others’ (1986/2004, p. 55). For example, he says (ibid., p. 54):

Now it is certainly true that Mary will acquire abilities of various kinds after herrelease. She will, for instance, be able to imagine what seeing red is like, be able toremember what it is like … But is it plausible that that is all she will acquire? … Onher release she sees a ripe tomato in normal conditions, and so has a sensation of red.Her first reaction is to say that she now knows more about the kind of experienceothers have when looking at ripe tomatoes.

THE EPISTEMOLOGICAL INTUITION AND THE ABILITY HYPOTHESIS 37


Jackson now embraces the ability hypothesis but it seems to me that his earlierline of argument is still plausible.

Even if the ability hypothesis can be defended against these objections, itsrole is only to provide a more nuanced version of type-A materialism, reduc-ing the counterintuitive impact of denying that Mary gains new propositionalknowledge on her release. It does not provide any strong, independent reasonsin favour of type-A materialism and against type-B materialism.21 Thus, nei-ther representationalism nor the ability hypothesis offers materials for thecritical agenda that any type-A materialist needs to develop. They do nothingto undermine the idea that there is more to know about human experiencethan is entailed a priori by the totality of physical facts.

1.11 Physical properties in new guises?If physicalism is true then phenomenal properties are physical properties.Specifically, if Jackson’s a priori physicalism is true then phenomenal proper-ties are physical properties either in the core sense (properties that figure in thephysical sciences) or in the extended sense (properties whose distribution isdetermined a priori by the distribution of physical properties in the core sense).If Jackson’s or Tye’s representationalism is true then we can say more aboutphenomenal properties: they are representational properties. The leading ideaof type-B materialism is that Nagelian subjective conceptions of types of expe-rience are conceptions of physical properties. The explanatory gap is consistentwith physicalism; there is habitable space between epistemological leaving outand metaphysical leaving out; there can be new knowledge that is not knowl-edge of new properties.

1.11.1 The ‘old fact, new guise’ response to the knowledgeargument

Type-B materialism is conceptually inflationist physicalism. It involves a dual-ity of objective and subjective conceptions without a metaphysical duality ofphysical and non-physical properties. The type-B materialist’s response to theknowledge argument is to accept the epistemological first premise but denythat this leads to the metaphysical conclusion that physicalism is false. WhenMary is released from her black-and-white room, her new knowledge involvesa new subjective conception of a fact that she already knew under an old


21 Yuri Cath (in press) argues that the ability hypothesis leads ultimately to the idea of newconceptions, and so turns out to be a version of the type-B materialist’s ‘old fact, newguise’ response to the knowledge argument (see below, section 1.11).


objective conception. As Levine says (1993, p. 125): ‘the case of Mary typifiesthe phenomenon of there being several distinguishable ways to gain epistemicaccess to the same fact’.

This is a common response to the knowledge argument (e.g. Horgan 1984),often known as the ‘old fact, new guise’ response (or the ‘old fact, new mode’response, to suggest Gottlob Frege’s (1892) notion of mode of presentation).Jackson rejects it, both early and late. When he first put forward the knowl-edge argument against physicalism, he already rejected the suggestion that theargument depends on ‘the intensionality of knowledge’ (1986/2004, p. 52).22

He now accepts physicalism and is convinced that the knowledge argument‘must go wrong’. But he still rejects the suggestion that the argument goeswrong in neglecting new conceptions, guises, or modes of presentation(Braddon-Mitchell and Jackson 2007, p. 137): ‘This is the explanation ofMary’s ignorance that is available to dual attribute theorists, not the explana-tion available to physicalists.’

There can certainly be multiple conceptions of the same physical property.But Jackson maintains that explaining Mary’s new knowledge by appeal tonew conceptions is incompatible with physicalism. In order to understandwhy, it is useful to recall the example of water and H2O (section 1.6.2). OnJackson’s view, it is a matter of conceptual analysis that water is the stuff thatfills the water role. Knowledge that water is H2O can only be arrived at a poste-riori. But the fact that water is H2O is entailed a priori by the fact that H2O fillsthe water role. Now suppose that some physical stuff, S, fills two roles, R1 andR2, in the physical order. Then we can have two conceptions of S. We can thinkof S as the stuff that fills role R1 or as the stuff that fills role R2. It may very wellbe that examining these conceptions themselves will not tell us that they aretwo conceptions of the same physical stuff. It is likely that this knowledge canonly be arrived at a posteriori. But if Mary, in her black-and-white room,knows the full story about the physical order, then she is already in a positionto know that the stuff that fills role R1 also fills role R2. This kind of exampleof multiple conceptions is available to a physicalist, but it does not provide a model for Mary’s gaining new knowledge on her release.

The situation would be different if S were to instantiate a non-physicalproperty, N. Then we could have a third conception of S. We could think ofS as the stuff that instantiates N. Even if Mary knew all there is to know about

PHYSICAL PROPERTIES IN NEW GUISES? 39

22 The intensionality of knowledge is illustrated by the fact that ‘Nigel knows that Hesperusis a planet’ may be true while ‘Nigel knows that Phosphorus is a planet’ is false eventhough Hesperus = Phosphorus.


the physical order, she might not be in a position to know that the stuffthat fills role R1 and role R2 also instantiates N. In her black-and-white room,she might know nothing at all of property N. But, while this kind of exampleof multiple conceptions would provide a model for Mary’s gaining newknowledge on her release, it is obviously not available to a physicalist (Jackson2005b, p. 262).

1.11.2 Descriptive and non-descriptive conceptionsAccording to the conceptually inflationist physicalist (type-B materialist),Mary gains new knowledge on her release because she gains new subjectiveconceptions of physical properties. The problem for type-B materialism is thatwe have not been able to find a model for Mary’s new conceptions that is con-sistent with physicalism.23

It is plausible that the source of this problem lies in the fact that we haveconsidered only conceptions that pick out a kind of physical stuff by description.These are conceptions of the form ‘the physical stuff that has property F’,where the property in question might be a physical property or a non-physicalproperty. Descriptive conceptions in which the descriptive property is physicalor physical-functional (such as ‘the physical stuff that fills the water role’) arealready available to Mary while she is in her room. So it may seem that new,distinctively subjective conceptions must be descriptive conceptions in whichthe descriptive property is non-physical (such as ‘the physical stuff that hasproperty N ’). Thus, if all conceptions are descriptive then Nagel’s duality of objective and subjective conceptions requires a metaphysical duality of physical and non-physical properties right from the outset. Since type-Bmaterialism proposes a duality of conceptions without a duality of properties,it requires that subjective conceptions—including the conceptions that become available to Mary only on her release—are not descriptive conceptions.

A partial analogy for the distinction between objective conceptions andnon-descriptive subjective conceptions is provided by the distinction betweentwo kinds of conception of locations in space. One kind of conception speci-fies locations in terms of distances and directions from an objective point oforigin. A location, L, might be specified as being 25 miles north-west fromCarfax. Deploying that conception of L in thought, I might achieve propositional


23 See Jackson (2005a, p. 318): ‘[T]he guises … must all be consistent with physicalism ifphysicalism is true…. But then, it seems, Mary could know about their applicability wheninside the room.’


knowledge (by looking at a map or reading a book, for example) that there iswater at L—perhaps the book says that there is a pond with ducks.A different kind of conception of a location is made available to me when I amat that location. Without knowing how far I am from Carfax or in whichdirection, I might arrive at a location and decide to explore a little. What isgoing on here? I notice sheep grazing on the other side of a stone wall, somefarm buildings further back, a tractor and, in the distance, trees. Then I see apond with ducks. So, there is water here.

If I am, in fact, 25 miles north-west from Carfax then this is an example ofa new instantiation of an old spatial property: I myself am at location L.In virtue of my new location, I gain new abilities: I can feed the ducks and, justby bending down, I can put my hand in the pond, I can see (and may laterremember) things that I have never seen before. But I do not only gain newabilities. I also gain a new indexical or egocentric conception of a location that I already knew about under a different, map-based, ‘distance and directionfrom origin’ or allocentric conception. I have a new cognitive capacity: I canthink of location L as ‘here’. Deploying the new conception in propositionalthinking, I achieve new propositional knowledge that (as I put it) ‘there iswater here’.

We should not rush from this partial analogy to the idea that subjective con-ceptions of types of experience are indexical conceptions, like thecontext-dependent conceptions of locations, times, and people expressed by‘here’, ‘now’, and ‘I’. In fact, there are reasons to reject the proposal that subjec-tive conceptions are indexical conceptions (Tye 1999; Papineau 2007). Onedisanalogy is that at least some subjective conceptions can be deployed inthought by a subject who is not concurrently having an experience of the typein question. They seem to function like recognitional, rather than indexical,conceptions (Loar 1997). But, in the face of Jackson’s objection to the ‘old fact,new guise’ response, even the partial analogy offers some encouragement tothe conceptually inflationist physicalist.

1.12 Phenomenal concepts and physicalismNon-descriptive subjective conceptions of phenomenal types of experienceare often called phenomenal concepts. Papineau introduces the idea with threemain points. First, he says (2002, p. 48): ‘when we use phenomenal concepts,we think of mental properties, not as items in the material world, but in termsof what they are like’. Second, he stresses that ‘as a materialist, I hold that evenphenomenal concepts refer to material properties’ (ibid.). Third, he insists thatthe advocate of phenomenal concepts must avoid the ‘poisoned chalice’ (p. 86)

PHENOMENAL CONCEPTS AND PHYSICALISM 41


of considering phenomenal concepts as descriptive concepts. Phenomenalconcepts refer ‘directly, and not via some description’ (p. 97).24

In an earlier and seminal paper, Brian Loar says (1997/2004, p. 219):

On a natural view of ourselves, we introspectively discriminate our own experiencesand thereby form conceptions of their qualities, both salient and subtle …. What weapparently discern are ways experiences differ and resemble each other with respect towhat it is like to have them. Following common usage, I will call these experientialresemblances phenomenal qualities, and the conceptions we have of them, phenomenalconcepts. Phenomenal concepts are formed ‘from one’s own case’.

Loar goes on to highlight the distinction between concepts and properties andto point to the possibility of accepting a duality of concepts or conceptionswithout a duality of properties (1997/2004, pp. 220–1):

It is my view that we can have it both ways. We may take the phenomenological intu-ition at face value, accepting introspective concepts and their conceptual irreducibility,and at the same time take phenomenal qualities to be identical with physical-func-tional properties of the sort envisaged by contemporary brain science. As I see it, thereis no persuasive philosophically articulated argument to the contrary.

1.12.1 A limitation on the promise of phenomenal conceptsPhenomenal concepts provide a model for subjective conceptions of types ofexperience—including the new conceptions that Mary gains on her release—and the model holds some promise of being consistent with physicalism. First,according to type-B materialism, subjective conceptions are conceptions ofphysical properties. Second, a phenomenal concept of a type of experience is anon-descriptive concept. It is a recognitional concept that a thinking subject pos-sesses in virtue of having an experience of the type in question. Deploying aphenomenal concept in thought is not a matter of thinking of a physical propertyas the property that has such-and-such higher-order property (that is, such-and-such property of properties). So the type-B materialist need not face an objectionalong the lines that phenomenal concepts can only account for new knowledge ifthey involve non-physical higher-order properties (section 1.11.2). Nevertheless,the promise of phenomenal concepts is limited in an important way.

According to physicalism, conscious mental states are physical states and the phenomenal properties of conscious mental states are physical properties.


24 In Thinking About Consciousness (2002), Papineau defends a ‘quotational’ or ‘quotational-indexical’ model of phenomenal concepts. More recently (2007), he acknowledges thatthis model faces some objections and he adopts a different view of phenomenal conceptsas cases of, or at least as similar to, perceptual concepts—something like ‘stored sensorytemplates’ (2007, p. 114). This change leaves intact the three points in the main text.


In general, instantiating a physical property is not sufficient for gaining a con-ception of that property but, according to type-B materialism, the phenomenalproperties of conscious mental states are special in this respect. Consider asubject who is, in general, able to form concepts and deploy them in thought.By being in a conscious mental state—having an experience of a particulartype—such a subject can gain conceptions of certain physical properties ofthat state, namely, the phenomenal properties of that type of experience.These conceptions are direct, non-descriptive, subjective, phenomenal con-cepts and, intuitively, the subject gains a phenomenal concept of a physicalproperty in such cases only because there is something that it is like to instan-tiate that property.

Now, recall Nagel’s remark (1974/1997, p. 524):

If mental processes are indeed physical processes, then there is something that it islike, intrinsically, to undergo certain physical processes. What it is for such a thing tobe the case remains a mystery.

According to Nagel, if there is something that it is like to instantiate certainphysical properties then we have no answer to the question why this is so.

We have just said that, intuitively, if we gain phenomenal concepts of certainphysical properties by instantiating them then there must be something that itis like to instantiate those properties. Consequently, it seems, if we gain phe-nomenal concepts of certain physical properties by instantiating them then,ultimately, we have no answer to the question why that is so. Possessing phe-nomenal concepts of physical properties does not have a fully satisfyingexplanation in physical terms.25

1.12.3 An argument against type-B materialism?This limitation on the promise of phenomenal concepts may offer the prospectof an argument against the type-B materialist’s claim that phenomenal con-cepts are direct, non-descriptive concepts of physical properties.

A subject who has a phenomenal concept of a type of experience meets therequirement of knowing which type of experience is in question (section 1.2.2).The subject knows what that type of experience is like (in one use of ‘knowingwhat it is like’) in virtue of being, or having been, the subject of an experienceof that type. But, a subject who knows which type of experience is in questionneed not think of that type of experience as the property with such-and-suchphysical-functional specification nor, indeed, as being a physical property at all.


25 For discussion of what can reasonably be demanded of the phenomenal concept responseto the knowledge argument, see Chalmers (2007), Levine (2007), Papineau (2007).


Furthermore, if Nagel is right then a subject who knows what a type of experi-ence is like has no answer to the question why this is what it is like, or why thereis anything at all that it is like, to instantiate a property whose nature is physical.

We might begin to wonder whether a subject can really possess a direct andnon-descriptive concept of a physical property, and meet the requirement ofknowing which physical property is in question, just in virtue of knowingwhat a particular type of experience is like. A subject who knows what a typeof experience is like has a phenomenal concept. But we might wonder whetherthinking about a physical property by deploying a phenomenal concept must beindirect, with the phenomenal concept embedded in a descriptive conceptalong the lines of: ‘the physical property that it is like this to instantiate’.26

Developing, and then responding to, these inchoate concerns would requirework on the metaphysics of properties—their individuation and theirnatures—and work on the ‘knowing which’ requirement that would inevitablylead into theories of reference in philosophy of language and thought. It is notobvious in advance what the outcome of this work would be. But suppose, fora moment, that it were to uncover a good argument for the claim that, if thereare distinctively subjective phenomenal concepts of phenomenal properties,then these phenomenal properties are not identical with physical properties(they neither are, nor are determined a priori by, properties that figure in thephysical sciences).

This would be an important argument. First, it would show that type-Bmaterialism can be rejected—that a duality of objective and subjective con-ceptions requires a duality of physical and non-physical properties—and itwould show this without simply relying on an assumption that all conceptionsare descriptive (section 1.11.2). Second, by showing that type-B materialismcan be rejected, the argument would provide the needed motivation for thesecond premise of the original knowledge argument (section 1.6.2) and—what comes to the same thing—it would license the transition from thelimited conclusion of the simplified knowledge argument, that type-A materi-alism is false, to the more sweeping conclusion of the original knowledgeargument, that physicalism is false (section 1.6.4).

Third, the argument would tie together the two requirements of Jackson’s a priori physicalism (section 1.8.3). The first requirement is that all propertiesshould be physical properties, defined as properties that are determined


26 Chalmers (1999) and Horgan and Tienson (2001) argue against the claim that direct phe-nomenal concepts are concepts of physical-functional properties. Also recall McGinn’scomment that ‘if we know the essence of consciousness by means of acquaintance, then wecan just see that consciousness is not reducible to neural or functional processes’ (2004, p. 9).


a priori by properties that figure in the physical sciences. The second require-ment is that all the facts should be entailed a priori by the physical facts.Suppose that the first requirement is met, so that physical properties are all theproperties there are. It would follow from the envisaged argument that thereare no distinctively subjective conceptions of any properties. But if all proper-ties are physical properties and there are no subjective conceptions, then thereis no impediment to a priori entailment of all the facts by the physical facts. Sothe second requirement would also be met (cf. Jackson 2005b, p. 260).

Fourth, the argument would figure as an item on the critical agenda that anytype-A materialist needs to develop. Jackson needs to undermine the intuitionthat there is more to know about human experience than is entailed a priori bythe totality of physical facts. While representationalism is consistent withphysicalism, it does not reveal any error or confusion in the epistemologicalintuition that drives the knowledge argument (section 1.10.1). Nor does theability hypothesis provide strong, independent reasons in favour of type-Amaterialism (section 1.10.2). By showing that type-B materialism can berejected, the argument would reveal that, even if type-A materialism involvessome cost to intuition, it is the only alternative to dualism.

1.12.4 Options for physicalismFollowing Chalmers, we have divided physicalist approaches to the philosophyof consciousness into two varieties, type-A materialism (also known as con-ceptually deflationist physicalism) and type-B materialism (also known asconceptually inflationist physicalism). We have just considered, in a specula-tive way, a possible line of argument against type-B materialism. If there wereto be a good argument of the envisaged kind then the options would seem tobe severely limited. A physicalist, having rejected the dualist options of inter-actionism and epiphenomenalism, would seem bound to embrace thecounterintuitive commitments of type-A materialism.

In fact, this is not quite right. At the beginning of this chapter, when I firstcontrasted dualism and physicalism (section 1.1.1), I said that, according tophysicalism, phenomenal properties are either identical with physical proper-ties or else strongly determined (necessitated) by physical properties. I alsosaid that the causal argument for physicalism allows for both a strict identityversion and a relaxed supervenience version of physicalism (section 1.7.2). Inrecent sections, however, I have adopted Jackson’s terminology. His version ofphysicalism says that all properties are physical properties. He allows that physicalproperties include properties that do not themselves figure in the physical sciences but are determined a priori by properties that do figure there. He doesnot allow that properties that are determined or necessitated only a posteriori



by physical properties are themselves physical. Some varieties of superve-nience physicalism are now classified as varieties of dualism and, specifically, asnecessitarian dual attribute theories.

This means that we need to reconsider the hypothetical situation if therewere to be a good argument against the type-B materialist’s claim that phe-nomenal concepts are direct, non-descriptive concepts of physical properties.If the dualist options of interactionism and epiphenomenalism were rejectedthere would still be two options available, not only type-A materialism, butalso the necessitarian dual attribute view. Theorists who describe this view as avariety of supervenience physicalism, rather than dualism, will regard it as con-ceptually inflationist, rather than deflationist, physicalism. As a consequence,it will be grouped with type-B materialism. But it will, apparently, be leftuntouched by the line of argument against type-B materialism that we consid-ered in section 1.12.3. According to the necessitarian dual attribute view,phenomenal concepts are not concepts of physical properties, but concepts ofdistinct phenomenal properties that supervene on physical properties.

The costs and benefits of the variety of supervenience physicalism alsoknown as the necessitarian dual attribute view are not, of course, affected by aterminological decision between ‘physicalism’ and ‘dualism’. In this chapter, wehave seen only one argument against this option and that was an Ockhamist27

argument in favour of the austerity of ‘bare physicalism’ (section 1.8.3). Thebenefits of austerity would have to be weighed against the costs to intuition oftype-A materialism.

The less austere, but otherwise more intuitive, option is favoured byEdmund Rolls (this volume), who leaves it as an open question whether it isbest described as ‘physicalism’. According to his higher-order syntactic thought(HOST) theory of consciousness, conscious mental states are physical states ofa system with a particular computational nature. The computational proper-ties of the state necessitate phenomenal properties a posteriori (p. 154; someemphases added):

[T]he present approach suggests that it just is a property of HOST computationalprocessing with the representations grounded in the world that it feels like something.There is to some extent an element of mystery about why it feels like something, why itis phenomenal … In terms of the physicalist debate, an important aspect of my pro-posal is that it is a necessary property of this type of (HOST) processing that it feelslike something… and given this view, then it is up to one to decide whether this view isconsistent with one’s particular view of physicalism or not.


27 The fourteenth-century philosopher, William of Ockham, is credited with a law of parsi-mony: ‘Entities should not be multiplied beyond necessity.’


1.13 ConclusionAt the first choice point in the philosophy of consciousness, some philoso-phers deny that there is an explanatory gap and accept type-A materialism. Wehave seen that Jackson joins Farrell and Dennett in this group, rejecting theintuition that there is more to know about human experience than what isentailed a priori by a battery of physical fact and theory that can be grasped byMary in her black-and-white room, or by a sufficiently intelligent Martian orbat. Philosophers who, at the first choice point, accept that there is an explana-tory gap proceed to a second choice point. There, some opt for dualism, othersfor type-B materialism.

Jackson’s knowledge argument and Chalmers’s conceivability argument are arguments for dualism. If these arguments are correct then the phenome-nal properties of experience are not physical properties. Philosophers whoaccept this conclusion—Jackson (at an earlier stage when he accepted theknowledge argument) and Chalmers (still)—face a further choice about thecausal relationship between the phenomenal and the physical. One option isto accept dualist interactionism at the cost of rejecting the completeness ofphysics. Another is to accept epiphenomenalism at the cost of rendering phe-nomenal properties ‘idle and beyond our ken’, as Jackson (2006, p. 227) nowputs it. These are not especially attractive views but Chalmers (2002) arguesthat these options, and others, should be taken seriously ‘if we have independ-ent reason to think that consciousness is irreducible’ (2002, p. 263).28

Chalmers also commends a view, Russellian or type-F monism (Russell1927), on which the most fundamental properties of the physical world areboth protophysical and protophenomenal—the physical and the phenomenalare variations on a common theme (2002, p. 265–6): ‘One could give the viewin its most general form the name panprotopsychism, with either protophe-nomenal or phenomenal properties underlying all of physical reality.’ Thisview is speculative and exotic, but Chalmers suggests that ‘it may ultimatelyprovide the best integration of the physical and the phenomenal within thenatural world’ (p. 267; see also Stoljar 2006).

According to type-B materialism, we can accept that consciousness is con-ceptually irreducible but reject dualism. This is an attractive option that isadopted by many contemporary philosophers of consciousness—probably themajority—including Block, Levine, Loar, Papineau and Tye. If type-B materi-alism can be defended then arguments for dualism are undermined and some

CONCLUSION 47

28 Chalmers (2002) refers to dualist interactionism as type-D dualism and to epiphenome-nalism as type-E dualism.


of the motivation for more exotic views, such as Russellian monism, isremoved. Indeed, the knowledge argument against physicalism and in favourof dualism seems to rest on the assumption that type-B materialism is not areal option for the physicalist.

There are, however, arguments against type-B materialism—against the ideathat we can have a duality of objective and subjective concepts, and anexplanatory gap, without a duality of physical and non-physical properties.Some of these arguments seem to depend on the assumption that all conceptsare descriptive and the dominant form of type-B materialism thereforeappeals to direct, non-descriptive, subjective, phenomenal concepts of physi-cal properties. But there are also arguments against this form of the view.

It may very well be that none of these arguments is, in the end, compelling andthat Loar (1997) will turn out to be right in saying that we can ‘have it both ways’:irreducibly subjective phenomenal concepts are nevertheless concepts of physicalproperties of the kinds that figure in neuroscience. On the other hand, there maybe a good argument against phenomenal concepts of physical properties andfriends of type-B materialist may have to consider shifting to the necessitariandual attribute view that phenomenal concepts are concepts of non-physical phenomenal properties that are determined or necessitated—but not a priori—by physical properties. Some philosophers may object to the departure fromontological austerity (Jackson 2003) and others may have concerns about a prim-itive relation of a posteriori necessitation between properties (strong necessities;see Chalmers 1996, 1999, 2002). But if, at the first choice point, there are goodreasons to accept that there is an explanatory gap then, at the second choicepoint, the necessitarian dual attribute view should be taken at least as seriously asRussellian monism, dualist interactionism, or epiphenomenalism.

AcknowledgementsI am grateful to Tim Bayne, Ned Block, Tyler Burge, Alex Byrne, DavidChalmers, Frank Jackson, David Papineau, Edmund Rolls, Nick Shea, DanielStoljor, and Larry Weiskrantz for comments and conversations.

ReferencesAlter, T. (2007). Does representationalism undermine the knowledge argument? In Alter, T.

and Walter, S. (eds) Phenomenal Concepts and Phenomenal Knowledge: New Essays onConsciousness and Physicalism, pp. 65–76. Oxford: Oxford University Press.

Armstrong, D.M. (1968). A Materialist Theory of the Mind. London: Routledge & Kegan Paul.

Armstrong, D.M. (1996). Qualia ain’t in the head. Review of Ten Problems of Consciousness:A Representational Theory of the Phenomenal Mind, by Michael Tye. Psyche 2(31);http://psyche.cs.monash.edu.au/v2/psyche-2-31-armstrong.html.



Block, N. (1978). Troubles with functionalism. In Wade Savage, C. (ed.) Perception andCognition: Issues in the Foundations of Psychology, Minnesota Studies in the Philosophy ofScience, Volume 9, pp. 261–325. Minneapolis: University of Minnesota Press. Reprintedin Block, N. (ed.) Readings in the Philosophy of Psychology, Volume 1, pp. 268–306.Cambridge, MA: Harvard University Press, 1980.

Block, N. (2002). The harder problem of consciousness. Journal of Philosophy 99, 391–425.

Block, N. (2003). Mental paint. In Hahn, M. and Ramberg, B. (eds) Reflections and Replies:Essays on the Philosophy of Tyler Burge, pp. 165–200. Cambridge, MA: MIT Press.

Block, N. (2005). Two neural correlates of consciousness. Trends in Cognitive Sciences9, 46–52.

Block, N. Flanagan, O., and Güzeldere, G. (eds) (1997). The Nature of Consciousness:Philosophical Debates. Cambridge, MA: MIT Press.

Braddon-Mitchell, D. and Jackson, F.C. (1996). Philosophy of Mind and Cognition: An Introduction. Oxford: Blackwell.

Braddon-Mitchell, D. and Jackson, F.C. (2007). Philosophy of Mind and Cognition: An Introduction, 2nd edn. Oxford: Blackwell.

Burge, T. (1997). Two kinds of consciousness. In Block, N., Flanagan, O., and Güzeldere, G.(eds) The Nature of Consciousness: Philosophical Debates, pp. 427–434. Cambridge, MA:MIT Press.

Burge, T. (2003). Qualia and intentional content: Reply to Block. In Hahn, M. andRamberg, B. (eds) Reflections and Replies: Essays on the Philosophy of Tyler Burge,pp. 405–616. Cambridge, MA: MIT Press.

Burge, T. (2007). Reflections on two kinds of consciousness. In Foundations of Mind: Essays by Tyler Burge, Volume 2, pp. 392–419. Oxford: Oxford University Press.

Byrne, A. (2001). Intentionalism defended. Philosophical Review 110, 199–240.

Byrne, A. (2006). Review of There’s Something About Mary, by Peter Ludlow, YujinNagasawa, and Daniel Stoljar. Notre Dame Philosophical Reviews (20.1.2006)http://ndpr.nd.edu/review.cfm?id=5561.

Cath, Y. (in press). The ability hypothesis and the new knowledge-how. Noûs.

Chalmers, D.J. (1995). Facing up to the problem of consciousness. Journal of ConsciousnessStudies, 2, 200–19. Reprinted as ‘The hard problem of consciousness’ (pp. 225–235) and‘Naturalistic dualism’ (pp. 359–368) in Velmans, M. and Schneider, S. (eds) TheBlackwell Companion to Consciousness. Oxford: Blackwell, 2007.

Chalmers, D.J. (1996). The Conscious Mind: In Search of a Fundamental Theory. Oxford:Oxford University Press.

Chalmers, D.J. (1999). Materialism and the metaphysics of modality. Philosophy andPhenomenological Research 59, 473–496.

Chalmers, D.J. (2000). What is a neural correlate of consciousness? In Metzinger, T. (ed.)Neural Correlates of Consciousness: Empirical and Conceptual Issues, pp. 17–39.Cambridge, MA: MIT Press.

Chalmers, D.J. (2002). Consciousness and its place in nature. In Chalmers, D.J. (ed.)Philosophy of Mind: Classical and Contemporary Readings, pp. 247–272. Oxford: OxfordUniversity Press.

Chalmers, D.J. (2007). Phenomenal concepts and the explanatory gap. In Alter, T. andWalter, S. (eds) Phenomenal Concepts and Phenomenal Knowledge: New Essays onConsciousness and Physicalism, pp. 167–194. Oxford: Oxford University Press.

REFERENCES 49


Davies, M. and Humberstone, I.L. (1980). Two notions of necessity. Philosophical Studies38, 1–30.

Dennett, D.C. (1988). Quining qualia. In Marcel, A.J. and Bisiach, E. (eds) Consciousness inContemporary Science, pp. 42–47. Oxford: Oxford University Press.

Dennett, D.C. (1991). Consciousness Explained. Boston: Little, Brown.

Dennett, D.C. (2005). Sweet Dreams: Philosophical Obstacles to a Science of Consciousness.Cambridge, MA: MIT Press.

Dennett, D.C. (2007). What RoboMary knows. In Alter, T. and Walter, S. (eds) PhenomenalConcepts and Phenomenal Knowledge: New Essays on Consciousness and Physicalism,pp. 15–31. Oxford: Oxford University Press.

Evans, G. (1982). The Varieties of Reference. Oxford: Oxford University Press.

Farrell, B.A. (1950). Experience. Mind 59, 170–198.

Frege, G. (1892). On sense and reference. In Geach, P. and Black, M. (eds) Translations fromthe Philosophical Writings of Gottlob Frege, pp. 56–78. Oxford: Blackwell, 1970.

Horgan, T. (1984). Jackson on physical information and qualia. Philosophical Quarterly 34,147–152. Reprinted in Ludlow, P., Nagasawa, Y., and Stoljar, D. (eds) There’s SomethingAbout Mary: Essays on Phenomenal Consciousness and Frank Jackson’s KnowledgeArgument, pp. 301–308. Cambridge, MA: MIT Press, 2004.

Horgan, T. and Tienson, J. (2001). Deconstructing new wave materialism. In Loewer, B.(ed.) Physicalism and Its Discontents, pp. 307–318. Cambridge: Cambridge UniversityPress.

Jackson, F.C. (1982). Epiphenomenal qualia. American Philosophical Quarterly 32, 127–36.Reprinted in Ludlow, P., Nagasawa, Y., and Stoljar, D. (eds) There’s Something AboutMary: Essays on Phenomenal Consciousness and Frank Jackson’s Knowledge Argument,pp. 9–50. Cambridge, MA: MIT Press, 2004.

Jackson, F.C. (1986). What Mary didn’t know. Journal of Philosophy 83, 291–295. Reprintedin Ludlow, P., Nagasawa, Y., and Stoljar, D. (eds) There’s Something About Mary: Essayson Phenomenal Consciousness and Frank Jackson’s Knowledge Argument, pp. 51–56.Cambridge, MA: MIT Press, 2004.

Jackson, F.C. (1995). Postscript. In Moser, P.K. and Trout, J.D. (eds) ContemporaryMaterialism, pp. 184–189. London: Routledge. Reprinted in Ludlow, P., Nagasawa, Y.,and Stoljar, D. (eds) There’s Something About Mary: Essays on Phenomenal Consciousnessand Frank Jackson’s Knowledge Argument, pp. 409–415. Cambridge, MA: MIT Press,2004.

Jackson, F.C. (1998a). From Metaphysics to Ethics: A Defence of Conceptual Analysis. Oxford:Oxford University Press.

Jackson, F.C. (1998b). Postscript on qualia. In Jackson, F.C. Mind, Method, andConditionals, pp. 76–79. London: Routledge. Reprinted in Ludlow, P., Nagasawa, Y.,and Stoljar, D. (eds) There’s Something About Mary: Essays on Phenomenal Consciousnessand Frank Jackson’s Knowledge Argument, pp. 417–420. Cambridge, MA: MIT Press,2004.

Jackson, F.C. (2003). Mind and illusion. In O’Hear, A. (ed.) Minds and Persons (RoyalInstitute of Philosophy Supplement 53), pp. 251–271. Cambridge: CambridgeUniversity Press. Reprinted in Ludlow, P., Nagasawa, Y., and Stoljar, D. (eds) There’sSomething About Mary: Essays on Phenomenal Consciousness and Frank Jackson’sKnowledge Argument, pp. 421–442. Cambridge, MA: MIT Press, 2004.



Jackson, F.C. (2004). Foreword. In Ludlow, P., Nagasawa, Y., and Stoljar, D. (eds) There’sSomething About Mary: Essays on Phenomenal Consciousness and Frank Jackson’sKnowledge Argument, pp. xv–xix. Cambridge, MA: MIT Press.

Jackson, F.C. (2005a). Consciousness. In Jackson, F.C. and Smith, M. (eds) The OxfordHandbook of Contemporary Philosophy, pp. 310–333. Oxford: Oxford University Press.

Jackson, F.C. (2005b). The case for a priori physicalism. In Nimtz, C. and Beckermann, A.(eds) Philosophy—Science—Scientific Philosophy: Main Lectures and Colloquia of GAP 5,Fifth International Congress of the Society for Analytical Philosophy, Bielefeld, 22–26September 2003, pp. 251–265. Paderborn: Mentis.

Jackson, F.C. (2006). On ensuring that physicalism is not a dual attribute theory in sheep’sclothing. Philosophical Studies 131, 227–49.

Jackson, F.C. (2007). The knowledge argument, diaphanousness, representationalism. Alter,T. and Walter, S. (eds) Phenomenal Concepts and Phenomenal Knowledge: New Essays onConsciousness and Physicalism, pp. 52–64. Oxford: Oxford University Press.

Kirk, R. (2006). Zombies. In Zalta, E.N. (ed.) The Stanford Encyclopedia of Philosophy(Winter 2006 edition), http://plato.stanford.edu/archives/win2006/entries/zombies/.

Kripke, S.A. (1980). Naming and Necessity. Cambridge, MA: Harvard University Press.

LeDoux, J.E. (1996). The Emotional Brain: The Mysterious Underpinnings of Emotional Life.New York: Simon and Schuster.

Levine, J. (1983). Materialism and qualia: The explanatory gap. Pacific PhilosophicalQuarterly 64, 354–361.

Levine, J. (1993). On leaving out what it’s like. In Davies, M. and Humphreys, G.W. (eds)Consciousness: Psychological and Philosophical Essays, pp. 121–136. Oxford: Blackwell.

Levine, J. (2007). Phenomenal concepts and the materialist constraint. In Alter, T. andWalter, S. (eds) Phenomenal Concepts and Phenomenal Knowledge: New Essays onConsciousness and Physicalism, pp. 145–166. Oxford: Oxford University Press.

Lewis, D. (1988). What experience teaches. Proceedings of the Russellian Society. Sydney:University of Sydney. Reprinted in Lycan, W.G. (ed.) Mind and Cognition: A Reader,pp. 499–518. Oxford: Blackwell, 1990.

Loar, B. (1997). Phenomenal states (revised version). In Block, N. Flanagan, O., andGüzeldere, G. (eds). The Nature of Consciousness: Philosophical Debates, pp. 597–616.Cambridge, MA: MIT Press. Reprinted in Ludlow, P., Nagasawa, Y. and Stoljar, D. (eds)There’s Something About Mary: Essays on Phenomenal Consciousness and Frank Jackson’sKnowledge Argument, pp. 219–239. Cambridge, MA: MIT Press, 2004.

Lodge, D. (2001). Thinks º London: Secker and Warburg.

Ludlow, P., Nagasawa, Y. and Stoljar, D. (eds) (2004). There’s Something About Mary: Essays on Phenomenal Consciousness and Frank Jackson’s Knowledge Argument.Cambridge, MA: MIT Press.

McGinn, C. (2004). Consciousness and Its Objects. Oxford: Oxford University Press.

McLaughlin, B.P. (2005). A priori versus a posteriori physicalism. In Nimtz, C. andBeckermann, A. (eds) Philosophy—Science—Scientific Philosophy: Main Lectures andColloquia of GAP 5, Fifth International Congress of the Society for Analytical Philosophy,Bielefeld, 22–26 September 2003, pp. 267–285. Paderborn: Mentis.

Merikle,P.M., Smilek, D., and Eastwood, J.D. (2001). Perception without awareness:Perspectives from cognitive psychology. Cognition 79, 115–134.

REFERENCES 51


Nagel, T. (1974). What is it like to be a bat? Philosophical Review 83, 435–450. Reprinted inNagel, T. Mortal Questions, pp. 165–180. Cambridge: Cambridge University Press, 1979.Also reprinted in Block, N. Flanagan, O., and Güzeldere, G. (eds) The Nature ofConsciousness: Philosophical Debates, pp. 519–527. Cambridge, MA: MIT Press, 1997.

Nemirow, L. (1980). Review of Mortal Questions by Thomas Nagel. Philosophical Review89, 473–477.

Nemirow, L. (2007). So this is what it’s like: A defense of the ability hypothesis. In Alter, T.and Walter, S. (eds) Phenomenal Concepts and Phenomenal Knowledge: New Essays onConsciousness and Physicalism, pp. 32–51. Oxford: Oxford University Press.

Nida-Rümelin, M. (1995). What Mary couldn’t know: Belief about phenomenal states.In Metzinger, T. (ed.) Conscious Experience, pp. 219–441. Paderborn: Mentis. Reprintedin Ludlow, P., Nagasawa, Y. and Stoljar, D. (eds) There’s Something About Mary: Essays onPhenomenal Consciousness and Frank Jackson’s Knowledge Argument, pp. 241–267.Cambridge, MA: MIT Press, 2004.

Nida-Rümelin, M. (2002). Qualia: The knowledge argument. In Zalta, E.N. (ed.) The Stanford Encyclopedia of Philosophy (Fall 2002 edition),http://plato.stanford.edu/archives/fall2002/entries/qualia-knowledge/.

Papineau, D. (2002). Thinking About Consciousness. Oxford: Oxford University Press.

Papineau, D. (2007). Phenomenal and perceptual concepts. In Alter, T. and Walter, S. (eds)Phenomenal Concepts and Phenomenal Knowledge: New Essays on Consciousness andPhysicalism, pp. 111–144. Oxford: Oxford University Press.

Quine, W.V.O. (1953). From a Logical Point of View. Cambridge, MA: Harvard UniversityPress.

Rosenthal, D.M. (1991). The independence of consciousness and sensory quality. InVillanueva, E. (ed.) Philosophical Issues, Volume 1: Consciousness, pp.15–36. Atascadero,CA: Ridgeview. Reprinted in Rosenthal, D.M. Consciousness and Mind, pp. 135–148.Oxford: Oxford University Press, 2005.

Rosenthal, D.M. (2005). Consciousness and Mind. Oxford: Oxford University Press.

Russell, B. (1927). The Analysis of Matter. London: Kegan Paul.

Ryle, G. (1949). The Concept of Mind. London: Hutchinson.

Ryle, G. (1954). Dilemmas: The Tarner Lectures 1953. Cambridge: Cambridge UniversityPress.

Seager, W. and Bourget, D. (2007). Representationalism about consciousness. In Velmans, M.and Schneider, S. (eds) The Blackwell Companion to Consciousness, pp. 261–276. Oxford:Blackwell.

Smart, J.C.C. (1959). Sensations and brain processes. Philosophical Review 68, 141–56.

Stoljar, D. (2001). Two conceptions of the physical. Philosophy and PhenomenologicalResearch 62, 253–281. Reprinted in Ludlow, P., Nagasawa, Y. and Stoljar, D. (eds) There’sSomething About Mary: Essays on Phenomenal Consciousness and Frank Jackson’sKnowledge Argument, pp. 309–331. Cambridge, MA: MIT Press, 2004.

Stoljar, D. (2004). The argument from diaphanousness. In Ezcurdia, M., Stainton, R. andViger, C. (eds) New Essays in the Philosophy of Language and Mind (SupplementaryVolume of the Canadian Journal of Philosophy), pp. 341–390. Calgary: University ofCalgary Press.

Stoljar, D. (2005). Physicalism and phenomenal concepts. Mind and Language 20, 469–494.



Stoljar, D. (2006). Ignorance and Imagination: The Epistemic Origin of the Problem ofConsciousness. Oxford: Oxford University Press.

Stoljar, D. and Nagasawa, Y. (2004). Introduction. In Ludlow, P., Nagasawa, Y. and Stoljar, D.(eds). There’s Something About Mary: Essays on Phenomenal Consciousness and FrankJackson’s Knowledge Argument, pp. 1–36. Cambridge, MA: MIT Press.

Sturgeon, S. (1994). The epistemic view of subjectivity. Journal of Philosophy 91, 221–235.

Tye, M. (1995). Ten Problems of Consciousness: A Representational Theory of the PhenomenalMind. Cambridge, MA: MIT Press.

Tye, M. (1999). Phenomenal consciousness: The explanatory gap as a cognitive illusion.Mind 108, 705–725. Reprinted in Tye, M. Consciousness, Color, and Content, pp. 21–42.Cambridge, MA: MIT Press, 2000.

Tye, M. (2000). Consciousness, Color, and Content. Cambridge, MA: MIT Press.

Velmans, M. and Schneider, S. (eds) (2007). The Blackwell Companion to Consciousness.Oxford: Blackwell.

Watkins, M. (1989). The knowledge argument against ‘the knowledge argument’. Analysis49, 158–160.

Weiskrantz, L. (1986). Blindsight: A Case Study and Implications. Oxford: Oxford UniversityPress.

Weiskrantz, L. (1997). Consciousness Lost and Found. Oxford: Oxford University Press.

REFERENCES 53


Consciousness and explanation - mkdavies.net · Consciousness and explanation ... tions and...

Documents

Transcript of Consciousness and explanation - mkdavies.net · Consciousness and explanation ... tions and...